ML20151Z109
| ML20151Z109 | |
| Person / Time | |
|---|---|
| Site: | 05200003 |
| Issue date: | 11/30/1997 |
| From: | WESTINGHOUSE ELECTRIC COMPANY, DIV OF CBS CORP. |
| To: | |
| Shared Package | |
| ML20138L015 | List: |
| References | |
| WCAP-14808-ERR, WCAP-14808-ERR-R03, WCAP-14808-ERR-R3, NUDOCS 9809210162 | |
| Download: ML20151Z109 (128) | |
Text
t.
=... _ -
4wJ '
...r 1
ENCLOSURE 2 -
WESTINGHOUSE LETTER DCP/NRC1410 AUGUST 13,1998
)
I' l
i i
i-I.
I i
- i..
i L.
9909210162 980813
{DR ADOCK 05200003 PDR a
i l
3785s. doc.
l
(' c.
t a
l l
1.7 Two Phase Flow Model l
To calculate the vapor and liquid flow rates, NOTRUMP solves a mixture momentum equation and y
uses constitutive relationships (referred to hem as a " drift flux model") to separate the mixture into its vapor and liquid components.
1.7.1 Model Description a)
Mixture momentum equation The one-dimensional mixture momentum equation for a vertical pipe can be written (Reference 1-3):
DG 1 BAp,Ud 1 BAp,U,2 dP P,t=-p,g y
(1.7-1)
+
+ _--
= _ _ _ _
j Bt A
8x A
8x dx A
where the subscript M denotes mixture, w denotes wall, U, denotes relative veloci., (ft/sec), and:
a(1 - n)p,p, (1.7-2a)
Pr "
Pu G = pyU, (1.7-2b) u where the subscript v denotes the vapor region and 1 denotes the liquid region.
The second and third terms on the left-hand side are the momentum flux terms due to area chance ned mixture velocity and density gradients, including the effects of relative velocity between the phases.
In the NOTRUMP application to AP600, these momentum flux terms are ignored. This assumption is evaluated in Section 1.7.5.
The wall friction term in two-phase flow (second term on the right-hand side) for flow qualities up to 90 percent is evaluated in NOTRUMP by assuming that the flow is all liquid, then applying a two-phase friction multiplier:
I
- "..I 1 2
(1.7-3)
,dz,,
A 2, D,, p, where f is the friction factor and D is the pipe diameter (ft.). The correlation used for two-phase hydrau!ic loss is discussed in Section 1.9. For flow qualities above 90 percent, a linear transition is assumed to all vapor flow (Section 2.16).
November 1997 i
o \\3934w\\3934w.NON;1b-121797 1,7 1 Rev.3
With the above simplifications and after integrating over a flowlink length Ax, the NOTRUMP momentum equation becomes:
DG" = - AP I ' fax 'Gu,2 -p 8 Ax (17-4) dt 2, D,
p, A more convenient form of the momentum equation for numerical solution is in terms of volumetric
- flow. Consider the phasic mass conservation equations:
- Bap, 1 BAap,U,, p, (1.7-Sa)
St A
dx (1.7-5b)
B(1 - a)p, + 1 BA(1 - a)p,U, = - pu at A
dx F# is the vapor generation rate per unit volume'(lb/ft.3/sec.).
Where Expand the phase conseivation equations as follows:
Ba Sp 3
pvt + "Y '+ p BAaU' + aU, p' = F" f gx (1.7-6a) x a
(1.7-6b) p' 3(1.a) + (1 - a)_StOp, + p, BA(1 - a)U, + (1 - a)U' p, = - 1*"
- at A
dx 8x a
Then, dividing by vapor and liquid density respectively, noting that the da/dt's cancel, assuming that i
the rate of change of phase densities is small, and adding results in the volumetric flow equation:
f 1 BAj, py 1 _1 (1.7 7)
A dx
, p, p,,
The types of transients to which NOTRUMP is applied involve relatively slow depressurization
- transients, with insignificant vapor superheating, so that the assumption of nearly uniform phase density is reasonable. It can be seen from Equation 1.7-7 that, if the vapor generation rate is also
. small, the volumetric flow is insensitive to vapor fraction gradients, unlike the mass flow rate. This means that the volumetric flux at a node boundary remains constant, even though the density at the boundary may undergo a discontinuous change. It is therefore advantageous to solve the momentum equation in terms of volumetric flow (see Section 2.4 for coding details). The approach taken and assumptions made are discussed after introduction of the drift flux model in the next section.
l November 1997 oA3934w\\3934w.NON:lb-121797 1,7 2 Rev.3 0
1 b) Drift flux n.odel
- Define 'he volumetric flux of vapor and liquif,, Qgf,ji (ft/sec.), and relative velocity U, in tegps7qfa) t
' the local velocities U, and U (ft/sec.):
i j, = (1 - a)U, (1.7-8b)
U, = U, - U, (1.7-8c)
. The vapor fraction a is the fraction of the local volume taken up by the vapor phase
' Also define the mixture volumetric flux j, as:
j = j, + J,
-(1.7-9)
Another relative quantity often used is the drift velocity, V,.
This is defined as the velocity of the vapor U, relative to the volumetric flux j. It can be shown using the above equations that:
V, = U, -j = (1 - - n)U,
' (1.7-10)
Rearrangement then gives the following basic equations of the drift flux model in terms of the drift
- velocity:
. j, = otj + aV (l.7-11a) g j, = (1 - a)j - nV (1.7-I lb) g
. The above equations are definitions that apply locally. To use these equations in real situations, they must'be cast in forms that apply over the entire pipe. Variations in local vapor fraction and mixture velocity across the pipe now become important, and complicate the drift flux equations. How this averaging and separating is done depends to some extent on the flow regime being considered. For example, Equations 1.7-Ila and 1.7-Ilb can be averaged as follows (Reference 1-4):
(j,) = (aj) + (aV )
- (1.7-12a) g (j,) = ((1-a)j) -(aV )
(1.7-12b) g Define a distribution parameter C, as:
Novunber 1997 oA3934w\\3934w.NON:Ib 121797 1,7 3
.Rev.3 J
<Gl>
C.713)
C* = <axj>
. Applying averaging to the second term of Equations 1.712a and 1.7-12b results in a weighted average drift velocity:
<aV > = <<x>
(1.7-14a) g
<a>
= <a>< <V,> >
= <a><(1 - a)U,>
Introducing Equations 1.7-13 and 1.7-14c into Equations 1.7-12a and 1.7-12b gives:
<j,> = C,<a> <j> + <a> <<V >>
(1.7-15a) g
- <j,> = (1 - C,<tx>)<j> - <a> <<V >>
(1.7-15b) g f
When the phases are highly separated, such as in annular flow, averaging across the pipe using Equation 1.7-13 leads to C,=1.0. In these situations, it is sometimes mere convenient to use an average drift velocity defined by:
<V > = (1 - <a>)<U,>
- (1.7-16) g The corresponding drift flux equations are:
<j,> = <a> <j> + <a> <V >.
(1.7-17a) g l
<j,> = (1 - <a>)<j> - <a> <V,>
(1.7-17b)
The expression for mass velocity in the pipe becomes:
G = (p, - Ap<a>) <j>. - op <a> <V,>
(1.7-18) y In the following sections, it is useful to compare the various drift flux expressions on the same basis.
. From Equations 1.7-15a/b and 1.7-17a/b, <V > can be related to <<V >> by:
g e
i I
I.
November 1997 0:\\3934w\\3934w.NON:Ib-121797 1.7-4 Rev.3 E
l
<Vp =
Vf> + (C, - 1)<j>
(1.7-19)
In the following text, the brackets around the fluxes and the vapor fraction are removed. The brackets l
around the two forms of the drift velocity are retained.
l i
From Equation 1.7-18, the rate of chang-of mass flow obtained through the momentum equation is
! recast in terms of the rate of change of volumetric flow (Section 2.4 contains additional coding
~
l
- details)f BApyUu
)DAj (1.7-20)
Bt Bt t
where transient changes in drift velocity, vapor fraction, and phase density aie assumed negligible compared to changes in volumetric flaw. Note that these conditions should be approximately true because the drift flux model relies on drift velocities derived from steady-state tests.
1.7.2 Constitutive Relationships
~
In the next subsections, the forms chosen for the drift velocity and C, are described. Since the drift flux model must be applied to upward and downward flow in vertical pipes, and also to flow m j
horizontal pipes, applicable correlations must be used.
In terms of geometry, there are two distinct regions in which two-phase flow must be modeled in the AP600. First, there are various pipes of both vertical and horizontal orientation, ranging from less i
than 1 in. in diameter to over 2 ft. Since these components contain the most significant hydraulic
- resistances, the highest mixture velocities are through these pipes.
Another group of components includes large diameter vessels in which pools of two-phase mixture reside. These are all at least several feet in diameter. Because the volumetric flow through the system is roughly conserved and is limited by the piping, fluid velocities in these regions are substantially smaller, and phase separation is more extensive.
The objective of the following subsections is to describe the basis for the drift flux models used in NOTRUMP for flow in vertical and horizontal pipes. Several simple models are described that, though not used explicitly in the NOTRUMP application to AP600, serve to clarify certain features and simplify assumptions of the drift flux model.
l
!a I
November 1997 c:d934w\\3934w.NON:lb-121797
],7-5 Rev.3
_)
1.7.3 Vertical Flow Models a) Low vapor fraction concurrent flow l
For a variety of low void fraction conditions, it is found that the drift velocity is a simple function of l
the phasic properties if the effects of variations in fluid conditions across the pipe are accounted for.
This is the case for a variety of flow regimes such as bubbly, slug, and churn turbulent flow (Reference 1-4). C,in Equation 1.7-13 varies from 1.1 to about 1.3, and <<Vp> can be expressed by two equations, one of which applies in small diameter pipes in which slug flow can occur, and one in larger diameter pipes where churn turbulent flow is likely:
.m 8 pD (1.7-21 a)
A
<<V*>> = 0.35 Pi W
0 8P (1.7-21b)
<<Vg>> = 1.53 Pi Several investigators have correlated vapor fraction in pools. For example, Sudo (Reference 1-5) expressed vapor fraction as a function of several property groups and vapor volumetric flux to some power, assuming that the liquid volumetric flux was negligible in comparison to the vapor flux. He found that the assumption of negligible liquid flux is reasonable for values up to about 1 ft/sec. The data range for this correlation is 1 to 100 atmospheres, and 0.2 to 1.5 ft. diameter. An earlier conelation, derived using similar assumptions, is the Yeh correlation (Reference 1-6), which is used in NOTRUMP. The Yeh correlation is based on rod bundle data, while the Sudo correlation is based primadly on pipes of various diameters. These conelations are compared in Figure 1.7-1. It can be seen _that the Yeh correlation agrees reasonably well at low to moderate vapor fractions with the more recent Sudo model. The lower vapor fraction at low values of j, is expected due to the open nature of the tube bundle, allowing for larger bubbles and larger drift velocities. - Although the Sudo correlation database does not extend to diameters typical of some of the AP600 components, extrapolation indicates lower vapor fraction for a given vapor flux, consistent with Yeh. The Sudo model indicates a likely change in flow regime for void fractions exceeding about 0.6 (the flat portion
. of the curve). Above a volumetric flux of about 20 ft/sec., the Yeh correlation may overpredict the vapor fraction.
These void fraction correlations, which are usually in the form a = Cjf, can be put in forms convenient for use in the drift flux model, 2s long as the vapor fraction is relatively low and the liquid flux is small compared to the vapor flux. Using Equation 1.7-17a, the following equation is obtained
. (Section 2.3 includes additional coding details):
November 1997 Rev.3 c:u934w'3934w.NON:lb.121797 1,7_6 b_
cx<V > = j, - uj g
7-22a)
<V,> -(1 - a)j/a (1.7-22b)
= ( I - ")"
(1.7-22c)
C '*
The EPRI model (Reference 1-7) uses the same basic form of the drift flux model expressed by Equation 1.7-13, but makes the variables <<Vgo and C, functions of the phase fluxes, fluid propedies, and hydraulic diameter, and then uses these functions to fit a variety of data. Because of its applicability over a wide range of conditions, the EPRI model is used in modified form in NOTRUMP for nearly all the flow paths in AP600 except for large open areas such as the vessel and CMT. The modifications to the EPRI correlation are based on a review of the model and comparison with the simpler models discussed above and are explained further in a subsequent section.
b) High vapor fraction coeurrent flow At high vapor fractions, the annular flow regime is likely to exist. A model based on an annular film (Reference 1-8) can be used.
Assume an annular film flowinF on the inside of a venical tube. The simplified momentum equations for each phase are:
'dP t,P, j + Pv8 " Au (1.7-23a)
'dP T..,P,,,
t,P, (1.7-23b)
,dx + p,g = A(1 - (x)
A(1 - a)
Where P, is the interfacial area per unit length (ft.2/ft.), P,i s the wall surface area in contact with the i
liquid (ft.2/ft.), t, is the interfacial shear stress (force per unit interfacial surface, Ibf/ft.') acting on the vapor, and t,,, is the wall shear stress acting on the liquid.
Subtracting one equation from the other elimings the pressure gradient tenn and results in:
t,,,P,,
t,Pi (1.7-24)
A(1 - a)
Acr(1 -a) t, can be expressed in a form similar to the wall shear stress, except that the relative velocity between the vapor and the liquid film is used:
November 1997 a:\\3934w\\3934w.NON:1tw121797 1,7 7 Rev.3 1
l f,3piU'i
- 3" (1.7-25a) 2 (1.7-25b) f,p,U,2 t
t' =
2 Wallis (Reference 19) developed an expression for the interfacial friction factor that is.widely used in these types of applications, and that visualizes an increasingly thick liquid film as equivalent to an r
increasingly rough pipe:
f,= 0.005[1 + 75(1 - a))
(1.7-26) i
' A solution of Equation 1.7.-24 to obtain the relative velocity was derived by Ishii (Reference 1-8).
{
The complete equations do not have a simple form, but 1shii simplified them to yield the following j
equation for <V >:
e I-"
gApD(1 - a)
(1.7-27)
<V > =
j+
g 3
0.015p, n+ 4/p/p, C, can be defined for this flow regime so that Equation 1.7-27 can be put in a form consistent with Equations 1.7-15a/b. C, and <<V >> for annular flow, using Equation 1.7-19 are:
g I-"
- C,,, = 1 +
a+ 4/p/p, (1.7-28a).
1 -a gApD(1 -- a)
(lil-28b)
<<V,,>> =
g a+ 4/p/p, S 0.015p, t
While the annular C, and <<V,>> are obtained differently from the low void fraction form described by Equations 1.7-13 and 1.7-14c, these terms are both describing the same thing: the slip of one phase relative to the other, in addition to the drift of vapor relative to the mixture as a whole.
The EPRI correlation fits functional forms of C, and <<V >> to a wide range of data in the high g
vapor fraction range, so this model is the preferred approach and is used in NOTRUMP. However, the annular model above is used to highlight certain common features of high vapor fraction flow and is also used tojustify certain simplifying assumptions.
L November 1997 oA3934w\\3934w.NON:lb-121797 1.7-8 Rev.3
c) Countercurrent flow' One of the most important flow regimes the drift Dux model must predict is the countercurrent regime.
' For situations where condensation is not significant, the most widely accepted flooding model, and that which is used by Westinghouse in safety analysis codes such as NOTRUMP, is based on'the Wallis flooding model.' The original form of the equation, due to Wallis, is written in dimensionless form as:
l j,"* + m(-ji )in = C i
(1.7-29a)
- lv.
3-
. j,..
J ApgD p,
(1.7-29b) p j,,/P/P,j ji y
J (1.7-29c)
ApgD Pi ApgD -
(1.7-29d) y, P,
. The characteristic velocity J (ft/sec.) is assumed to be a function of the tube diameter D. As the tube diameter increases, the liquid volumetric flux that can flow downward against a given. upward flux of vapor increases. This type of behavior is usually called J' scaling. The constant C is found to range from 0.7 to 1.0, and the constant m from 0.8 to 1 (Reference 1-10).
1 Tests at larger scales show that the appropriate length scale for large tubes is not the tube diameter but i
the Taylor instability wave length:
0 A.
(1.7-30)
-S Apg
= When the diameter length scale in J above is replaced by the Taylor length scale, the following characteristic velocity results:
nu UAPE (1.7-31)
K=
2 P,
'This characteristic. velocity is called the Kutateladze number, and the CCFL behavior under these conditions is called K* scaling. The nondimensional fluxes,' obtained by replacing J with K in November 1997 oA3934w\\3934w.NON:Ib-121797
},7 9 Rev.3 J
Equation 1.7 20d, are described as k*. In this case, the liquid downward flux for a given vapor flux
. remains unchanged as tube diameter increases. Ibr steam / water mixtures ranging from 15 to 1
1000 psia, K* scaling should apply for tubes larger than about 2 to 3 in, in diameter.
l i
'Ihe constant C changes' with tube geometry at large scale. However, wide variations in geometry do -
I not strongly affect the value of C. For flooding through holes in a plate, for example, tests by Bankoff (Reference 1-11) show that the value of C approaches two for large holes and thick plates l
typical of those in a PWR.-
t i
The constant m has been shown to be primarily a function of end conditions and to vary from 0.65 to 7
0.8 (Reference 1-11).
i In summary, the CCFL data without condensation at small scales (pipe diameter less than 2 in.) can be adequately represented by the following equation and range of constants.
r s ix.
j)"+ (0.8%1.0) $
(- j,)in = (0.7-+ 1.0) J in (1.7-32)
Pv, i
s i
For large pipes and orifices typical of those in a reactor, the data can be represented by the following equation:
3iu jj" + (0.7--+ 1.0) S
(- j3)38 = (1.5-+2.0)K in (l.7-33)
< P.,
.When plotted on the (j,)'8 and (ji)ti2 plane, the above equations are straight lines and define the boundary between permitted countercurrent flow and forbidden countercurrent flow. A two-phase flow computer model, be it drift flux or two fluid, should not predict counterturrent flow in the forbidden region.
i Several approaches can be taken to derive a CCFL mooel applicable at all scales (Reference 1-12).
The NOTRUMP vertical CCFL model uses the [
]" so that a transition j
takes place between J' and K* scaling as the pipe diameter increases. The constants m and C are 0.7, j
1.0 (for small diameter), and 1.79 (for large diameter), respectively, based primarily on large pipe flooding data (Reference 1-13). - Equations 1.7-32 and 1.7-33 are used to compare the NOTRUMP
. flooding predictions using simple pipe models of various diameters (Section 3.2).
In the following, the special forms of avg >> and C, are determined, which cause the drift flux model described by Equations 1.7-15a/b to sweep out the Wallis flooding curve as ot is varied, as shown in Figure 1.7-2. In this way, application of the drift flux model leads naturally to the correct limiting rates of countercurrent flow.
November 1997 j
o:\\3934w\\3934w.NoN:IbJ121797 1,7 10 Rev.3 1
.- - -... _ -. _.. - -. _. - - ~ -. - -
Starting with Equations 1.7-15alb, and letting a'=aC, and e ;<Vg>> = J,, eliminate j to yield:
C (1 - a')J,- - a'j,= J, (1.7-34)
Use Equation 1.7-33 to describe the flooding curve, and rewrite Equations 1.7-33 and 1.7 34 as:
. j,'"+ (- Mji)'8 - K,'* = 0 (1.7-35a) -
(1 - n')j, - n'j3 - J, = 0 _
(1.7-35b) where:
Pi M,= (0.7)2 (1.7-36a) '
$ p, (1.7-36b)
K,= (l.79)2K It is relatively straightforward'to show (Reference 1-14) that for the drift flux lines to be tangent to the CCi1 curve as a is varied, J, must have the form:
a'(1 - a')K*
J* =
(1.7-37) n'+ M,(1 - n')
A relationship must still be found between n' and 'a. It is observed (Reference 1-13) in large-scale
~
flooding tests that the point of zero liquid downflow (ji = 0) is defined by a characteristic vapor velocity and is independent of the vapor fraction (at high vapor fractions). This characteristic velocity, which also defm' es the constant K, in Equation 1.7-35a, is found to be:
l f
M l/4 08 4 (1,7 38)
K = 3.2 e
2-y g
j I
i i
3e 4
November 1997
~ o:0934wu934w.NON.lb-121797 1,7 11 Rev.3 1
a
Setting j, to zero in Equation' 1.7-35b, and using Equation 1.7 37:
J* -
a'K*
(1.7-39) i j,(j,=0) =
=
1 - a' ' a'+ M,(1 - a')
J l
! Since j, = aU,, and, at zero liquid penetration,' U, = K,:
i ct'K*
i j,(j,= 0)= aK,=
- (1.7-40) a'+ M,(1 - n')
i
- Thus:
I a=
(1.7-41) n'+ M,(1 - n')
. A multiplier can be defined, called C,,. Using Equations 1.7-37 and 1.7-41:
ot' = C,,a (1.7-42a) f l
L
~
(1.7-42b) e I-
-i L
i n.c (1.7-42c) h i
!~
Equations 1.7-42b and 1.7-42c describe a drift flux model that sweeps out the CCFL curve as illustrated in Figure 1.7-2 and, at high vapor fraction, predicts a vapor velocity at the liquid holdup i
- point consistent with'the Kutateladze number. The C, defined by Equation 1.7-42b is derived in a t
manner different from those defined by Equation 1.7-13 or Equation 1.7-28a. Here, it is simply r f
[
form needed to obtain the proper asymptotic behavior of the drift flux lines at the liquid hold up point.
3 L
' Figure 1.7-3 compares C., (Equation 1.7-28a) and C., as functions of vapor fraction. It can be seen o
that the behavior is similar, even though C,, applies to cocurrent flow. This similarity is taken l
. advantage of when the overall model is summarized.
E l
l-i
.I November 1997 i
a:u9Mwu9Mw NON:lb 121797 1.7-12 Rev.3 l
l l
i l
d) Application of EPRI drift flux model The EPRI drift Dux model is used in modified form for all vertical pipes in the RCS, as described below.
The drift velocity and distribution parameters derived from the EPRI correlation, [
]" are compared to the flooding model values, [
]" from Equations 1.7-42b/c, and the [
]" of each are taken. This forces the countercurrent flooding behavior to be bounded by the CCFL curve, as described by Equations 1.7-32 and 1.7-33. Although the EPRI correlation has provision for calculating countercurrent flow, the approach is more complicated, and the CCFL curve is based only on flooding data through orifice plates. Examination of the drift flux l
lines produced by the EPRI correlation indicates that a larger countercunent region is permitted by the unmodified EPRI model. The modified model therefore tends to produce more liquid holdup.
l 1.7.4 Horizontal Flow Models a) Cocurrent flow Several methods have been used to correct homogeneous flow theory for slip effects in horD.,ntal two-phase flow, or high mass velocity cocurrent vertical flows. These methods usually involve flow parameters or a slip ratio to relate flow variables such as flow quality or average volumetric flux to the average vapor fraction. The slip ratio is defined as:
<U,>
(1 - <a>)<j'>
(1.7-43)
S=
=
<Up
<a><j,>
Thom (Reference 1-15) found that the slip ratio was basically a function of the vapor and liquid density ratio, over a wide range of conditions. The table below summarizes Thom's data.
Slip Ratio in Two-Phase Mixtures Determined by Thom Density Ratio (p, / p,)
Slip Ratio 10 1.5 100 2.7 500 4.2 1000 5.5 i
November 1997 l
0:\\3934w\\3934w.NON:lb-121797
],7 13 Rev.3 l
1 Others (Reference 1-16) have used the flow parameter K, defined as:
9=K,',
(1.7-44)
Jv+Ji Typical values of K are on the order of 0.8.-
Both the slip ratio and the flow parameter can be expressed in terms of a horizontal C.h,' defined o
' below:
jy= C.haj (1.7-45a) o ji= (1 - C.ha)j (1.7-45b) o Substituting Equation 1.7-43 into Equation 1.7-45a:
3, (I -"No.h"3 (l.7-46) i a(1 - C.h")3 o
Rearranging, C becomes:
o (1.7-47)
C o'h = 1 + (S - 1)a From Equation 1.7-44, K = 1/C,3 o
' De values of C,n and C,3 calculated from Thom's data and from the flow parameter K are plotted in o
o Figure 1.7-4 and are compared with the low vapor fraction vertical flow form C,e (C,e = 1.2 to 1.3) o o
and the high vapor fraction form, C,, (Equation 1.7-28a) at atmospheric pressure, it can be seen that o
E he value of C.B from the Bankoff model (1/K = 1.25) is similar to that obtained for low vapor t
o fraction vertical flow, and the C.h btained from the Thom data is similar to the annular flow model, o
C ;, (and from Figure 1.7-3, also to the CCFL model, C,f).
o His similarity suggests that both vertical and horizontal concurrent flows can be described with the same model for C. For low vapor fraction flows, a constant value of 1.25 would apply; for annular o
or stratified flows, either C,, or C.f would apply. When the pipe is in the vertical orientation, the o
o same C models can be used; only the applicable V >> term needs to be added. This is what is o
g
- done in the NOTRUMP code.
1 November 1997
_ c:09Mwu9Mw.la.NON410998
],7 14 Rev.3 L
-)
l l'
b) Stratified / countercurrent flow An important flow regime in horizontal channels is the stratified flow regime. In this flow regime, steam can flow countercurrent to the liquid, and there is little interaction between the phases. An equally important aspect of these flows is the point at which the stratified flow regime transitions to a slug or bubbly regime. When the transition occurs, the interfacial drag between liquid and vapor increases significantly, and the phases are forced to move concurrently.
l The flow regime transition from stratified to slug, or plug flow was investigated for square channels l
by Wallis and Dobson (Reference 1-17). ney proposed the following criterion:
l 3U Iv = 0. Sot (3,7,4g,).
3v i
ApgD (1.7-48b)
S Pv A more complete model for transitions in horizontal flow in circular channels was developed by Dukler and Taitel (Reference 1-18). An expression similar to that obtained by Wallis was developed fo+ me stratified to intermittent boundary, with the constant replaced by a function of the stratified '
water level.
l l
As pointed out by Takeuchi (Reference 1-19), the stratified to intermittent boundary defined by the _
l above equations is also equivalent to Wallis's solution to the problem of wave stability in horizontal
)
channel flow (Reference 1-9). His solution defines a region of permissible concunent or countercurrent stratified flow in the jy, ja plane in which waves on the water surface remain stable.
I j
This regica is illustrated in Figure 1.7-5. - Outside this region, waves become unstable and cause the -
flow regime to change. Wallis's fonn of the equation for the wave stability problem is:
L l
jl14j*1G,3 (1.7-49)
For the Duckler Taitel flow regime transition, the equation is similar in form, but with different i
exponents:
jlI/3.j al/3, 3 (1.7-50)
L
--The characteristic velocity is assumed to be a function of the channel diameter D and is equivalent to J
{
in Equation 1.7-29d. In contrast to vertical flow, this relationship is expected to hold at both small and large scales. This is because interfacial disturbances on a horizontal interface are more stable than on a vertical interface, due to the additional stabilizing effect of gravity in the horizontal case.
November 1997 i/
oA39MwC9Mw.1tNON-121797 1,7-1 $
Rev.3
L For the Wallis-Dobson transition, the form is again similar but with different constants and exponents:
j " +ji
=.707 (1.7-51) y j-
' Equations 1.7-49 to 1.7-51 can be used to represent the stratified flow regime transition boundary.
l Since Equations 1.7-50 and 1.7-51 more accurately represent the data trends, these are used as benchmarks against which to compare predictions by the NOTRUMP code in Section 3.3.
It should be noted that the limits described by the above equations are stability limits, and as such should be viewed as upper limits to the stratified flow regime. The extent of countercuurrent flow is controlled primarily by the level gradients occurring in the pipe, as described below.
l l-The driving force for horizontal countercurrent flow is assumed to be an initial liquid level difference j
in the channel, as illustrated in Figure 1.7-11. Positive flow is defined from left to right. The liquid i
experiences an additional force due to the level difference Ah, which is:
- A ApgAh = - A(1 -a)ApgAh (1.7-52) i A balance is achieved when the interfacial drag is sufficient to offset this force. Integrating Equations 1.7-23a and 1.7-23b and assuming that wall shear is negligible results in:
T P;Ax i
-(AP + pygAk)
(1.7-53a)
=
A tP;Ax (1.7-53b)
-(AP+ pgAh)= - (1 - a)A This leads to:
T P Ax ii ApgAh =
(1.7-54) ot(1 - a)A The interfacial perimeter P (ft.) is approximately proportional to 4a(1-a)D, reaching a maximum i
when the pipe is half full, and going to zero at each extreme. Using Equation 1.7-26 for the interfacial friction factor and rearranging leads to:
For a wide range of vapor fractions at low pressure, the first term on the right hand side is approximately 0.6.
November 1997 oMWuMWur la.NON-121797 1.7 16 Rev.3 L
l l
' l/2 (1 - u) K AP8(Ah/Ax)D (1.7 55)
(1 - a)U' =..04[1 + 75(1 - (x)],
3 p,
The simple m'odel developed above indicates that the countercurrent flow drift velocity is likely to be substantially lower than the limit established by stability limits, unless the level gradient is on the order of 1.0.
l As the mixture flow increases, wall shear is likely to become important. In this case, a slip component enters via C, whose form is similar to those obtained for annular flow. To simulate the I
o overall behavior of horizontal flow, the following forms of Vd and C, are used:
i l
'a.c (1.7-56a) i l
' a.c i
(1.7-56b) i j
l-where:
P_1 M ** =
$ py (1.7-57a) i I
l (1.7-57b)
AP8(Ah/Ax)D K.h =
c Pv i
I As indicated in the previous subsection and demonstrated in Figures 1.7-2 and 1.7-3, the above expression for [
]*** represents concurrent stratified and annular flows reasonably well, while the Vg drift term accounts for countercurrent stratified flow at low mixture flow rates. These forms also result in drift flux lines that sweep out a parabolic countercurrent flow boundary, consistent with the stability limits (although these boundaries are well below the limits defined by Equations 1.7-49,
' 1.7-50, and 1.7-51).
a I-I+
November 1997 oM9Mw\\3934w.la.NON.121797
},7 17 Rev.3
1 1
1.7.5 Effect of Neglecting Momentum Flux Terms This section provides detailed justification for not including momentum flux in the NOTRUMP models.
NOTRUMP does not include the second and third terms of Equation 1.7-1. The momentum flux terms arise from the acceleration imposed on the flow by density and area changes along the pipe.
This can more easily be seen by simplifying the momentum equation to consider steady, horizontal, homogeneous equilibrium flow (the importance of drift or slip effects will be discussed later). The pressure gradient in this case is given by:
1,7-58
$ o. -
(W u, + W uf)
=-
g r
where P is the pressure, f is the friction factor, D is the pipe diameter, W is the mixture mass flow rate (subscripts f and g define the gas and liquid flows), A is the area, vf s the fluid specific volume i
(1/ density), $ is the two phase multiplier, and u is the mixture velocity. Assuming homogeneous (u = u ) conditions (the importance of drift or slip effects will be discussed later), the momentum r
g equation becomes:
i dP IVf 'W'2 2 _ W du 1.7-59 E,_2D
,A, 3
A dz Since u = Wy/A, where v is the mixture specific volume, the velocity derivative can be replaced to yield:
2 dP fV G $3o2-G 2 dv + G v dA 1.7 60 f
2
=-
_g Where G = W/A. The above equation separates the influence of changing specific volume, and changing flow area. In two phase flow, the mixture specific volume is given by:
l v = vg + xvfg l
l
[
where x is the steam quality. Since the phasic specific volumes are functions of pressure only, the l
spatial derivative can be split into:
l l'
l November 1997 o:u934wu934w.la.NON 121797
],7 18 Rev.3
- - -.. ~
l L
l l
l I
l r
dv dvg dvrg dP dx dx 1.7-61 l
dz
, dP dP,-dz + vfg dz = (xv,8 + (1 - x)v,g) _dP + vfg _dz
_=
+x dz where v' nmans the derivative of specific volume with respect to pressure. Combining the above equations pelds the following equation (similar to equation 2.44 of Reference 1-21:
fvf 2 dx ~ XK 1.7-62 v dA 2
G
+Y dP, '%
f8E dz 1+G2 [xyg (3_x)y )
f This~ equation shows that the momentum flux terms influence the pressure gradient locally (via the area and quality gradients in the numerator), and globally via the denominator.
The area change term is accounted for in the NOTRUMP momentum equation by adding an overall loss factor K to the frictional term which calculates the overall loss in pressure across the area change.
Typically, area changes in the AP600 piping network are abrupt, and therefore introduce additional
)
irrecoverable losses which must also be accounted for in the loss factor. Application of the momentum flux term shown above evaluates only the recoverable pressure change, and would not be accurate if applied without the additional or offsetting irrecoverable losses.
The quality gradient term is important where there is boiling due to heat transfer or flashing due to a pressure gradient. If the quality gradient is dominated by boiling from a surface, the energy equation gives:
fs) q /
f 1.7-63 W
dz where h denotes fluid enthalpy and q'is the local linear hea'. rate (Btu /ft/s). Expanding the derivative (assuming pressure effects are small), the acceleration effect due to boiling becomes:
L November 1997
, o:0934wu934w.la NON 121797 1.7-19 Rev.3
1 l
b=Vfg (q '/A) 1.7-64 yf8 sz h
G fg The momentum flux term may be important during the natural circulation period, where mixture density differences drive the flow. Estimates using typical values indicate this term is comparable to the friction term at low mass velocities in a boiling channel. In NOTRUMP, this component of the overall pressure drop is accounted for via the two phase multiplier correlation, which is derived from a data base which includes data in heated tubes (Reference 1-22, page 57).
NOTRUMP does not account for the increase in the overall pressure gradient resulting from the denominator in Equation 1.7-62. This term is examined in more detail below.
The increase in fluid quality can also be dominated by the pressure gradient. Across an orifice, for example, the fluid enthalpy can be assumed to be fairly constant, such that x = (h-h )/his where h is f
constant. Therefore:
dx 8
'h-h '
dP
-{ (h,f + xh,fg) dP V
f fg 1.7-65 Vfg E "Vfg y g
The quality gradient term in this case appears in the denominator such that:
~ fvf 2 dP,
%3 ~ y dA 2
1.7-66 A dz, dz 1+G2 [x(v'g - vfg h'g / h ) + (I ~ *}(V'f ~ Vfg h'f /h)]
fs fg where as before the h' terms denote derivatives with respect to pressure. The value of the denominator is controlled by mass velocity. At high mass velocities, the denominator becomes less than one and the pressure gradient is increased. The denominator in fact is a measure of the degree to which the flow is approaching critical conditions.
Effect of uneaual phase velocity Equation 1.7-62 and 1.7-66 assume that the liquid and vapor move at the same velocity. Most flows will develop some difference in phasic velocity. The effect of this relative velocity on the pressure gradient will be examined. Because the denominator is important at high mass velocities, and the slip 1
November 1997 oA3934w\\3934w.la.NON-121797
},7 20 Rev.3
~.
l ratio (u /ug) is a more appropriate measure of the flow condition than the relative velocity (u -ug), the g
g effect of unequal phase velocities will be examined using the slip ratio. 'Ihe mass velocity can be expressed as (Reference 1-21, Equation 3.17):
{
L
' -1
- 8 + (I ~")*f 1.7-67 o.
ti 's ug i
l Let S = u /ug. Then:
s
,3-
' -l G= 1 N
8 + (1 - x)vg
=
ut, S
cg r
r j
i i
r 3)
G=
[xy + S(1 - x)vg]-1 =
g ss s
r i
. Now evaluate the inenia term in Equation 1.7-58 assuming constant area and flow to simplify the derivation, and use the equations above to get:
2 d (W ug g + W uf) = G
[xc, + (1 - x)cf]
1.7-69 g
The expression'in square brackets can be rearranged to give:
xc, + (1 - x)ct = xv, + (1 - x)vg + x(1 - x)
-1 vg + (S - 1)vg 1.7-70 s
i-I Assundng the quality and slip ratio are independent of pressure, the denominator in Equation 1.7-62 is now:
l.-
November 1997 c:0934wV934w.la.NON-121797
],7 21 Rev.3
i 2
1.7-71 1+G y,g (j _ x)y,g + x(1-x)
-I v', + (S - 1)v',
The effect of the additional terms in Equation 1.7-71 relative to the denominator in Equation 1.7-62 will be examined below.
Calculated mass velocities in AP600 Figures 1.7-7 to 1.7-13 of Reference 1-20 show the vapor and liquid volumetric flux calculated by NOTRUMP in various components of the AP600. These figures indicate that mass velocities are generally quite low except in the ADS lines. It should be noted that the velocity shown for the ADS 4 -
line is higher than actually calculated in NOTRUMP, because the area used to calculate the velocity from the code output volumetric flow rate was about 20 percent smaller than was actually utilized in the NOTRUMP calculation. Nevertheless, fluid velocities are likely to be on the order of several hundreds of feet per second in these lines. Since the mass velocity is likely to be highest in the ADS lines when the valves are open and critical flow exists at the valves, the importance of the denominator in determining the overall pressure drop will be examined at these locations. For simplicity, the compressibility terms in Equations 1.7-62 and 71 will be evaluated; sample calculations indicate the quality terms in Equations 1.7-66 are smaller because of the relatively large value of h -
rg Table 1.7-2 lists fluid saturation properties at various pressures, from which the derivatives are obtained as shown. Table 1.7-3 evaluates the denominator at several pressures and qualities for ADSI-3, and Table 1.7-4 does the same for ADS 4 at the lower pressure. For each quality, the mass velocity in the piping approaching the valve is estimated by multiplying the critical mass velocity for the given pressure and quality (using the HEM model) by the area ratio of the valve to the upstream piping. These calculations indicate the following:
a)
For ADSl-3, the pressure gradient leading up to the ADS valves could be tmderestimated by as much as 9 percent (the denominator ranges from.91 to.99). However, during the important period of low quality two phase flow, when the mixture level is at the top of the pressurizer, the error is substantially smaller. By itself, this difference is not sufficient to explain the differences between predicted and measured mass flow rates which are observed in some of the OSU tests (see response to RAI 440.721(c)).
b)
For ADS 4, where two. valves are assumed open, Table 1.7-5 indicates that the pressure gradient could be significantly underpredicted during the initial period just after the. valves open, when the pressure is high enough that critical conditions exist. This is because the total valve area is comparable to the upstream piping area.
i Figure 1.7-6 shows the effect on the denominator of assuming a slip ratio of 6 for the conditions shown'in Table 1.7-4. It can be seen that phase slip increases the value of the denominator, and j
j reduces the effect of acceleration on the pressure gradient. This is consistent with the observation that L
November 1997 I
o:\\3934w\\3934w la.NON-121797
],7 22 Rev.3 j
l A
- _. _ _.._ __._ _.. _ ___ __ _._ _.._.___ _ _ _.-._ _ _.m._.
t the Moody critical flow model, which assumes a large slip ratio, predicts a higher critical mass velocity than HEM (i.e., acceleration must be greater to produce large pressure gradients and choked
. conditions). Therefore,~ the conditions estimated in Tables 1.7-3 and 4 for homogeneous flow are the most severe to be expected.
i NOTRUMP model for ADS niping and critical flow Figure 1.7-7 illustrates the noding used to model the ADS piping and valves in AP600. The piping from the hot leg or pressurizer to the valve is simulated with a fluid node. A portion of the overall line resistance is allocated to the flow link connected to the pressurizer or hot leg. he local static pressure and enthalpy in the ADS piping node, P and h, are used in the Henry-Fauske and HEM o
o l
critical flow models to calculate the critical mass velocity (Section 2.17 Reference 1-20). With this modeling, the frictional pressure drop in the piping leading to the ADS valve is accounted for. The HEM model is applied over the short remaining distance to the valve, where the effect of friction can I
be ignored. However, the NOTRUMP model contains two deficiencies:
a)
The model does not account for acceleration effects in calculating the pressure distribution up j
L to the ADS valve (previous sections).
l' b)
The model does not account for the effect of significant upstream kinetic energy on the critical j
flow calculation.
l As indicated in the previous section, lack of momentum flux terms in the momentum equation may result in an underprediction of the pressure d", to the ADS valves. In the next section, the effect of ignoring the kinetic energy terms in the calculation of critical flow is examined.
j De HEM critical flow model assumes frictionless adiabatic, steady flow and begins with the following simplified mass, energy and momentum conservation equations:
i-l dW = 0 l
l-2 l
d h +."_.
=0 l
2, s
J i
dP + pudu = 0 1.7-72 1-~
where h is the fluid enthalpy. Because the flow is assumed frictionless and adiabatic, the flow is isentropic. Therefore, either the momentum equation or the energy equation can be replaced by:
7 i
l.
November 1997
~ A3934w\\3934w.1a NON-121797
],7 23 Rev.3 l
o
.._.~_
~ _., _. _.. _ __
ds = 0 In the HEM, the energy and entropy equations are used. The differentials are expanded to give:
.2 2
u, uo 1.7-73 h,+ 7 =ho+7 s = s0 g
h where the subscript t represents the conditions at the throat, and the subscript 0 represents conditions at the location where the acceleration to the throat is assumed to begin.. Usually this is taken as a location where the kinetic energy is negligible (u is small). Given the stagnation enthalpy and o
entropy, the stagnation pressure and the conditions at the throat leading to the maximum mass velocity can be determined. In NOTRUMP, the Henry-Fauske and HEM models consist of a series of tables giving critical mass flux as a function of stagnation enthalpy, and stagnation pressure.
In the modeling of the ADS, the effect of a significant kinetic energy component at the start of the process must be examined. To determine what the appropriate stagnation pressure should be, retain the second form of the momentum equation, and expand the differential to yield:
2 2
P dP u,
-uo 1.7-74 0
P, p
2 Assume that an average density can be defined such that:
P, dP, P, o
1.7-75 o
P p
Then:
-2
-2 p
P"t
,p
. P"o 1.7-76 2
2 his indicates that the " reservoir" pressure should include the recoverable portion of the fluid dynamic pressure at the point where acceleration is to begin.
November 1997 o:u934wu934w.la.NON.121797 1.7-24 Rev.3
l l
1 I
l i
l Because of the energies and pressures involved, a significant velocity must exist at point 0 before j
significant error is introduced. For example, at 50 psia the enthalpy of steam is 1174 Btu /lb. For a 1 percent increase in the total enthalpy, the fluid velocity must be about 770 ft/s. For a 1 percent increase in the total pressure, a fluid velocity of about 200 ft/s is needed. This would indicate that the effect of including the dynamic pressure in the reservoir conditions is more important than the effect of including the kinetic energy. Ignoring these terms, as is done in NOTRUMP, would be expected to result in a prediction of critical Dow which is too low.
l To confirm this, an alternate flow calculation was performed on the ADS 4 piping system to compare with the NOTRUMP prediction (as noted previously, the effects of compressibility were determined to be most important for this component). For steam flow in a piping system, the effects of compressibility can be taken into account by the use of net expansion factors Y (Reference 1-23.
These factors are functions of the pressure difference through the pipe, and the loss coefficient in the l
pipe (Figure 1.7-8). The How rate through the pipe is calculated by the following equation (Equation 1-11, Reference 1-23:
2
@HL - P,) gut Iw Wgo4 = 0.525Yd s
K where d is the pipe diameter in inches. The calculated How rate through both valves assuming compressible conditions is compared with the incompressible result (Y=1) in Figure 1.7-9. To compare with the NOTRUMP AP600 predictions, vapor flow is plotted against hot leg pressure for the ADS 4 pipe in Figure 1.7-10. The NOTRUMP values are seen to remain below the calculated value assuming compressible conditions.
==
Conclusion:==
It is concluded that NOTRUMP has a' compensating error in regions where the fluid acceleration is significant. On the one hand, lack of a momentum Dux term causes the pressure gradient upstream of the valve to be underestimated. On the other hand, neglecting the dynamic pressure terms in the critical now model will tend to underestimate the critical flow rate. Both errors become significan only in the ADS 4 piping where both valves are open. The overall effect is to produce a low estimate of the vapor flow through ADS 4, as indicated in Figure 1.7-10. Since this will reduce the depressurization rate of the system and delay the onset of IRWST, the presence of these compensating
[
errors in NOTRUMP is judged to be acceptable.
l November 1997 c:U9Mwu9Mw.ltNON-121797
],7-25 Rev.3
-e Table 1.71 FIGURES DEPICTING RESULTS FROM VERTICAL'AND HORIZONTAL FLOW MODELS Figure No.
Title 1.7-1 Yeh Correlation versus Sudo Correlation 1.7 CCFL Curve and Tangent Drift Flux Lines 1.7-3 Comparison of Various Forms of C, for Vertical Flow Regime 1.7 Comparison of C, for Horizontal and Vertical Flow t
1.7-5 Flow Regime Transitions in Horizontal Flow 1.7-6 '
Effect of Slip Ratio on Momentum Flux Equation Denominator l;
1.7 NOTRUMP Noding for ADS Valves 1.7-8 Effect of Compressibility on the Calculated Flowrate Through a Piping System 1.7-9 Effect of Compressibility on ADS 4 Flow vs. Pressure 1.7 10 NOTRUMP Predicted Vapor Flow Compared with Reference 1-23 Calculation 1.7-11 Horizontal Stratified Flow with Level Gradient 3
i 1.7 12 Flow Regimes in Hot Leg and Cold Leg (High Pressure, Prior to ADS Opening) 1.7 13' Flow Regimes in Hot Leg and Cold Leg (Intermediate Pressure, Just After ADS Opening) 1.7 14 Flow Regimes in Hot Leg and Cold Leg (Low Pressure).
~1.7-15 Flow Regimes in Balance Line (High Pressure, Just Prior to ADS Opening) 1.7-16 Flow Regimes in Balance Line and Pressurizer Surge Line (Intermediate Pressure, Just L
After ADS Opening) 1.7 17 Flow Regimes in Balance Line and ADS-4 (Low Pressure)
-1.7-18 Flow Regimes in Steam Generator Tubes (High Pressure, Just Prior to ADS Opening) l.
11 i
I l
t 1
l l
L November 1997 o: 3934w 3934w-ItNON-121797 1,7 26 Rev.3 y
e'
-t:++r
---w a
eep----
- - - =
-r--
-- m
__.m
._.....---_._m
.m_...-_.______._
Table 1.7 2 WATER SATURATION PROPERTIES AND DERIVATIVES PRESS TSAT VF VG VFG HF HG HFG VF VG 40 267.25 0.017151 10.496 10.47885 236.15 1169.8 933.65 0
0 45
.274.44 0.017214 9.3988 9.381586 243.52 1172 928.48 1.26E-05
-0.21944 I
50 281.02 0.017274 8.514 8.4 % 726 250.25 1174.1 923.85 1.2E-05
-0.176 %
55 287.08 0.017329 7.785 7.767671 256.4 1175.9 919.5 1.lE-05
-0.1458 60 292.71 0.017383 7.1736 7.156217 262.2 1177.6 915.4 1.08E-05
-0.12228 i
90
. 320.28 0.017659 4.8953 4.877641 290.7 1185.3 894.6 0
0 95 324.13 0.0177 4.6514 4.6337 294.7 1186.2 891.5 8.2E-05
-0.04878 100 327.82 0.01774 4.431
' 4.41326 298.5 1187.2 888.7 8E-06
-0.04408 110 334.79
' O.01782 4.0484 4.03058 305.8 1188.9 883.1 8E-06
-0.03826 l
i l
120
'341.27 0.01789 3.7275 3.70961 312.6 1190.4 877.8 7E-06
-0.03209 420 449.4 0.01942 1.1057 1.08628 429.6 1204.7 775.1 0
0 1
460 458.5
.0.01959 1.0092 0.98961 439.8 1204.8 765 4.25E-06
-0.00241 500 467.01 0.01975-0.9276 0.90785 449.5 1204.7 755.2 4E-06
-0.00204 540 475.01 0.0199 0.8577 0.8378 458.7 1204.4 745.7 3.75E-06
-0.00175 580 482.57 0.02006 0.7971 0.77704 467.5 1203.9 736.4 4E-06
-0.00152 600 486.2 0.02013 0.76975 0.74962 471.7 1203.7 732 3.5E-06
-0.00137 l
620 489.74 0.02021 0.74408 0.72387 475.8 1203.4 727.6 4E-06
-0.00128 980 542.14 0.02152 0.4561 0.43458 539.5 1193.7 654.2 0
0
~1020 546.99 0.02166 0.4362 0.41454 545.6 1192.2 646.6 3.5E-06 0.0005
.1000 544.58 0.02159 0.44596 0.42437 542.6 1192.9 650.3 3.5E-06
-0.00049 1060 551.7 0.02181 0.4177 0.39589 551.6 1190.7 639.1-3.75E-06
-0.00046
-1100 556.28 0.02195 0.4006 0.37865 557.5 1189.1 631.6 3.5E-06
-0.00043 l-November 1997 oA3934wu934w.la.NON.121797
},7 27 Rev.3
I Table 1.7 3 ACCELERATION EFFEC 1'S IN ADS 13 PIPING (Note: See Table 1.7 5 for nomenclature)
PRESS =
50 AVALVE 0.324 APIPE =
0.6827 l
VF=
1.2E45 VFG =
8.496726 VF =
0.017274 VG' = ~
-0.17696 IIFG =
923.85 SLIP =
6 X
GCRIT GRIPE '
V 1+G2V' STERM 1+G2VS i
0.01 861 409
-0.00176 0.94 0.0015
- 0. 4 0.1 283 134
-0.01769 0.93 0.0133 0.9' O.5 145 69
-0.08847 0.91 0.0369 0.95 0.9 110 52
-0.15926 0.91 0.0133 0.91 1
0.99 106 50
-0.17519 0.91 0.0015 0.91 f
PRESS =
100 VF=
8E.06 VFG =
4.41326 VF =
0.01774 VG' =
-0.04408 HFG =
888.7 SLIP =
6 X
GCRIT GRIPE V
1+G2V' S*ERM 1+G2VS 0.01 1565 743
-0.00043 0.95 0.0004 0.99 l
0.1 539 256
-0.0044 0.94 0.0033 0.98 0.5 282 134
-0.02204 0.91 0.0092 0.95 0.9 216 103
-0.03967 0.91 0.0033 0.92 0.99 207 98
-0.04364 0.91 0.0004 0.91 PRESS =
600
_VF' =
3.5E-06 VFG =
0.74962 VF =
0.02013 l
VG' =
-0.00137 HFG =
732 SLIP =
6 X
GCRIT GRIPE V
1+G2V' STERM 1+G2V3 0.01 6284 2982
-lE-05 0.98 0.0000 1.00 0.1 2661 1263
-0.00013 0.95 0.0001 0.99 0.5 1599 759
-0.00068 0.92 0.0003 0.95 0.9 1254 595
-0.00123 0.91 0.0001 0.91 0.99 1204 571
-0.00135 0.90 0.0000 0.91 l
PRESS =
1000 VF' =
3.5E-06 VFG =
0.42437 VF =
0.02159 1
November 1997 o \\3934w\\3934w-la.NON-121797 1.7-28 Rev.3
Table 1.7 3 (cont.)
ACCELERATION EFFECTS IN ADSI 3 PIPING (Note: See Table 1.7 5 for nomenclature) i VG' =
-0.00049 11FG =
650.3 SLIP =
6 X.
GCRIT
. GRIPE V
1+G2V' STERM 1+G2V'S 0.01 8633 4097
-1.4E-M 0.99 4.2E-06 1.01 0.1 4177 1983
-4.6E-05 0.96 3.82E-05 0.99 0.5 2639-1252
-0.00024 0.92 0.000106 0.95 0.9 2101-997
-0.00044 0.91 3.82E-05 0.91 0.99 2021 959
-0.00048 0.90 4.2E-06 0.91
)
i l
l i
i November 1997 c:0934w\\3934w-la.NON-121797 1.7-29 Rev.3
Table 1.7-4 ACCELERATION EFFECTS IN ADSI 3 PIPING (Note: See Table 1.7 5 for nomenclature)
VF =
3.5E-06 VFG =
0.42437 VF =
0.02159 VG' =
-0.00049 HFG =
650.3 SLIP =
6 X
GCRIT -
GRIPE V
1+G2V' STERM 1+G2VS 0.01 8633 4097
-1.4E-06 0.99 4.3E-06 1.01 0.1 4177 1983
' 4.6E-05
'0.96 3.82E-05 0.99 0.5 2639 1252
-0.00024 0.92 0.000106 0.95 0.9 2101 997
-0.00044 0.91 3.82E-05 0.91 0.9 2101 997
-0.00044 0.91 3.82E-05 0.91 0.99 2021 959
-0.00048 0.90 4.3E-%
0.91 l
l l
l l-l.
f I
o:\\3934w\\3934w-ItNON 121797 1.7-30 ev.3 L
Table 1.7 5 ACCELERATION EFFECTS IN ADS 4 PIPING PRESS =
50 AVALVE 0.527 APIPE =
0.559 VF* =
1.2E VFG =
8.496726 VF =
0.017274 VG' =
-0.17696 IIFG =
923.85 SLIP =
6 X.
GCRIT GRIPE V
.1+G2V' STERM 1+G2V'S 0.01 861 812
-0.00176 0.75 0.0015 0.96 0.1 283 267
-0.01769 0.73 0.0133 0.93 0.5 145 136
-0.08847 0.64 0.0369 0.79 0.9 110 104
-0.15926 0.63 0.0133 0.66 0.99 106 50
-0.17519 0.63 0.0015 0.63
]
~ Nomenclature:
j 1
GCRIT
=
critical mass flux at ADS valve GRIPE mass flux upstream of the valve = GCRIT*AVALVF1APIPE
=
V'
=
x vg' + (1-x) vj 2
1+G V' denominator in equation 440.721(h) - 5
=
STERM last term in equation 440.721(h) - 13
=
2 1+G V'S equation 440.721(h)-14
=
i l,-
i November 1997
[
oA3934wu934w-la.NON-121797 1.7-31 Rev.3 I
l
I l
I i
i l
i 1.7.6 Conclusion To determine whether the NOTRUMP drift flux model is being applied within its range of applicability, flow regime maps were generated using the models of Taitel and Dukler (References 1-18 and 1-24). On j
these maps were placed calculated liquid and vapor volumetric fluxes from selected components from a F'TRINP calculation of the AP600 2-in. cold leg break. The results are shown in Figures 1.7-11 to 1;. 4. It can be seen that in general, horizontal pipes are always stratified, while vertical pipes traverse several of the flow regimes discussed in this section. The pressurizer surge line is a special case because it is a slanted, curved pipe. Tests in inclined pipes (Reference 1-25) indicate that the rtratified flow l
regime cannot be maintained in slightly inclined pipes, reverting instead to flow regimes more typical of vertical pipes. The surge line flowlink is therefore modeled as a vertical flowlink, and phasic flows are placed on the vertical flow regime map (see Figure 1.7-16). It can be seen that the expected flow regime j
is annular in the surge line. Because of the special geometry of the surge line, the NOTRUMP drift flux model may have difficulty predicting the correct phasic flows in this component.
It is concluded that the NOTRUMP drift flux model is applicable in both vertical and horizontal pipes.
The surge'line may present problems to the model due to its unique geometry. There are also indications that the model predicts excessive liquid holdup (i.e., a restrictive CCFL) relative to data in vertical pipes.
I J
l I
1 1
0:\\3934w\\3934w-la.NON.121797 1.7-32 v.
i
)
Oa e*
- c e
y Comparison of Yeh to Sudo Void Fraction Correlation; Atmospheric Pressure 5O?
0.9
. G G$
5 0.8 eo
,e e
w 0.7 f----
9
- 'p#
M g.
0.e.
n o
--- Sudo (D=.5 ft) c n
.e 0.5
,s g
--y
,e
- - - - - -Sudo (D=1 ft) y
'.... ~~ '
_. _. Sudo (D=1.5 ft)
O g
j E
.4 g
,/,.,
w 0
ye 4.*.,-?.,e mC
- 1..'
- y 0.3 17 1
l}
o Ij t
ne 02 -
b=
0.1
't 0
c<a 0
5 10 15 20 25 30 5
E Volumetric Flux (ft/s) 00 "
o-
- <W W
WW
l Flooding lines in countercurrent
-10.0 flow regime J
l
- 8.0
.- 6.0 r
l O,
/
/
- 4.0
,s',/
- 2.0
-d.0 -$.5 -$.0 -d.5 -d.0 -1.5 -1.0
.5 O!O O!5 1!0 d
JF Figure 1.7-2 CCFL Curve and Tangent Drift Flux Lines November 1997 i
OA3934wV934w-la NON-121797 1.7-34 Rev.3
,liil!
l 1I 1\\
l; il I
1 s
' 9 0
L E
8 D) 0 O (2 f,
7 MO
' 0 C
G ON I
6 CD 0
N O
O O
IT FL C
F.
OS_
5 A
0 R
V F
)
R N
(1 O
4 P
O o,
0 AV O
SC I
3 RR 0
A.
L AU PNN.
2 M
A.
0 O
1 C
I 0
0 2
9 8
7 6
5 4
3 2
1 9
1 1
1 1
1 1
1 1
1 1
0
! 2 g w p 4" h. % $ E a m* E % O I t h a 3 $ n f $
o
{3x sw hGE,i6g
- fd wa. w
5 VERTICAL VS HORIZONTAL CO MODELS d
a
.f i
E 2
6 5
1.9 -
.g g
2 5
1.8 -
L o
1.7 -
g
~
1 f
1.6 a
1.5 y
,n a
h m
1.4 g
Ck 8
1.3
\\
F
\\\\
g, 1.2 4
5 1.1 P
I I
I I
h 1
8 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
g A
B X
VAPOR' FRACTION
$2 L$
1: ANNULAR 2:. THOM 3: BANKOFF
- 4: CHURN
<e I
,...a
\\
l d
1,*
COUNTER CURRENT CO-CURRENT
/
-~
s
(
i
~ ;
/
I, f
%a Figure 1.7-5 Flow Regime Transitions in Horizontal Flow November 1997 o:0934wu934w.la.NON-121797 1.7-37 Rev.3
4 4
i i
2 A
i
)
1 t
EFFECT OF SLIP RATIO ON DENOMINATOR iM 4
0 00 0 80 0.70 0.60 5
h 0.50
--- HOMOGE5EdbS
- 0.40 -
SLIP =6 0.30 0 20 0 10-
. 0.00 0
0.2 0.4 0.6 0.8 1
QUALffY Figure 1.7-6 Effect of Slip Ratio on Momentum Flux Equation Denominator November i997 oM934w\\3934w-la.NON 121797
},7 38 Rev.3
i i
l 1
l I
l l
1 i
P.. h.
Ge i
l PRE l
I Ge P h.
l not Lao i
i 1
1 1
i 1
Figure 1.7-7 NOTRUMP Noding for ADS Valves November 1997
. o:\\3934w\\3934w-la.NON-121797
],7 39 Rev.3
4 i
i Not Expenslen Pector Y for Compressible Mew Threveh Pipe to e Larger Mew Aree l
l k = 1.3 k..rre.. mitts t I fue CO, 50, HQ H,$ \\He N,0. C!, CH, C He and C He' LO w l
u m m as Festeen
\\\\
CJS For Senis Veleewy
& = 1.3 M
3$$%
N
~
h NhN3llll 1.1
.sts
.6 3 NN](bNNb A 1.s
.sas
.ut Y l, TSSsMk f 2.'o"s NWSSSSb*(
N,*ie.h55%
\\
\\
N e
.rse
.no Y,,-#,fot is
.m
.res
(
(
(
0.85 is
.ser
.rts 3
3
\\ +*
5 to
.ut
.rts OR
- ~.
- N to
.str
.rts
+
f,.,l tse
.910
.71s 0
0.1 L2 0.3 0.4 L5 L6 0.1 0J 0.9 LO bP N
1 l
l 1
Figure 1.7-8 Effect of Compressibility on the Calculated Flowrate Through a Piping System November 1997 o:u9.uwu934w.la.NON-121797 1,7 40 Rev.3
l l
t ADS 4 FLOW VS PRESSURE USING COMPRES$1BILITY FACTOR 80.000 70 000 so W -
50.000
/-
5 R 40.000
~ ~ ~ "#
3 W4INC
,s#',
A s
h a
30.000
_e s'
/
/
20.000 - - -
j,4 s'
/
/
i 10.000
~
0.000 15 20 25 30 38 40 45 50 HOT LEO PRESSURE (PSIA) l l
Figure 1.7-9 Effect of Compressibility on ADS 4 Flow vs. Pressure November 1997 l
o:0934wu934w-la.NON.121797 1,7 41 Rev.3 i
4 AP.600 2lNBREAK IN FNODE 49 12 node core Moss Flow Rote (ibm /s) a wCFL 185 0
0 ADS 4A VAPOUA FLOW Moss Flow Rote (Ibm /s)
Blending value 80 80 s
E E
a a
- 60 60 -
~
e
~
40 40 J z.
b
, g&
a a
a 20 20 w w
W3 K
O s'
0 O
0 10 20 30 40 50 Pressuie (psia) l l
l l
l Figure 1.7-10 NOTRUMP Predicted Vapor Flow Compared with Reference 1-23 Calculation Novernber 1997 i
c:U934wu934w la.NON 121797
},7 42 Rev.3 l
l
[
t l
l 1.
4 l
l l-I
.U Ah y
=
If 6
. -=
=
1
~
K' i
1 l
Figure 1.711 Horizontal Stratified Flow with Level Gradient
{
November 1997 o:\\3934w\\3934w-la.NON.121797 1.7-43 Rev.3
l a
9 t
r g
i
(
t DISPERSED i
G 9
5
\\
~
tt 5
o r
100 a
n
=
-h 5
6 x
>5 m
ee en :E 4
E INTERMnTENT
-y)se E
10 r
=
=e E
e=
m E
==
>o
-v
=
o g
3 o
.R g,
P. o 8
A b o a
1 r
o
=
2 STRATIFIED ANNULAR r
y r
2 g
=-
' ^ ' ' ' ' '
i
?
0.1 n
2 m
1 10 100 1,000 0
=
l
.a Vapor Volumetric Flux (ft/s)
K
?
W" 5*
Pressure =1000 110:
Cold Q,G
[
D=2 teg Ug w$
o
n
""5 g-K r-g g -
a
.I I
~
DISPERSED 7
w 2
05
'M o
=
=*
a 5z
- e. g 100 i
=
x g
3 E
E
,C 8
INTERMITTENT 10
=
-i E
i 9
i
-a E
h.
C 3
n a
=
.E
=#
a a
4 1 -, = '
E ANNULAR
=
-n 7
a STRATIFIED b
0.1 E
1 10 100 1,000 2
l g
Vapor Volumetric Flux (ft/s)
X yM Pressure =500 Hot Cold j
Qg h
N2 kg hg w$
o
O sr E.
r DISPERSED
?
w 2
e 3
6 E,
a3 1
100 r
9
&m
-u
=
v s
x o
a 5
t x
6.
E INTERMITTENT T
10 r
[
E
=
F 2
>o i
y a
r 5
A e
ANNULAR v
5 a,
a n
1 Ir.
"a.
s i
=
S
~1r. . *' #.. = s.b
.."g l
==
i C'
- e "e =
3
- 1. " -
'.'4 ",. 1
't o
STRATIFIED g
5
..O 7
",' I
..O, i g
0.1 m
g i
1 10 100 1,000 g
Vapor Volumetric Flux (ft/s)
K
{
Pressure =20 lha Cold
<e D=2 Leg Irg L'S t
O
i a
l 100 i DISPERSED BUBBLE yg-
?
R n
o 9
s c;
^
f 5
m l
10 7
^
b F
e.
x 3
m a
g F
.o F
2:e E
l
=
a
=
~
r r
>C
~
F 4
o
.?,
A
=
so y
w
.-4 BUBBLE CHURN ANNULAR
?
E 0.1 r
E a
b Cn m
- 2.
l ir 0.01 g
>g 0.01 0.1 1
10 100 1
y hpor Volumetric Flux (ft/s)
F'.
au-e-iom s.w.
.g e
n=i m.
wm p
E 5
E, 100 r r
a i
i DISPERSED BUBBLE 8
V5 h
ni
/
10 r 5
of@
o 2a 4%
^
>=
o ea
" s-x kE d
I F
IE o
en
'E e
s
=
c 2
0.1
=
a E
w A
7, e
.3 m
.7
=
BUBBLE CHURN ANNULAR I
0.01
=
m E
t
- 1. g E
0.001 r
=
3"?
=
a w
t!
el tl
,l t
i i t i t i
e i t t t a t t 1
T t
t t t t I
t t
t t t t E
g E
0.01 0.1 1
10 100
=
g Vapor Volumetric Flux (ft/s)
R
?
h surgetine N ""*" "
Bd'"
D=1 une
<w c
.'"t
- c a
o uw
~~.
- k N
i
~
~
o og 0
0 0
0 o
o o
oO d
8 E
o A
o+
~H O
m
~
E u
c i k.
W o
D 2
=
A x
>o A
U
~
"e A
O 7
~
a
- il' IZ A
Q E
m m
M N
=
m w
~
~
Q J
o m
m f
A i
l' li,ii.,i 1,,,,,
li,ii,,
I,,,,,,,
y i
i i
8 8
y 5
~
o 4
(S/U) xnig ap;amn[oA P!nb g Figure 1.7-17 Flow Regimes in Balance Line and ADS-4 (Low Pressure)
I-November 1997 o11934w\\3934w-la.NON 121797 1.7 49 Rev.3
E E
.r E
p g"
100 E
z
~
8.
?
a 4
3 x
s e
es E'
10 7
^
2
=
x m
=
a E
a uc c
n R
E 1
=
2 2
Ei 8
f a
m u
e g
=
i.
.a_
ANNULAR W
BUBBLE g
sr 0.1 stuS w
ie m
e E
3 c
c n
0.01 j
0.01 0.1 1
10 100 m
o Vapor Volumetric Flux (Ns) t l
g SG Tubes i
D=.05 g
m x
o E
'O G
g i
w i
w~
_.3 1
1.8 Mixture Level Tracking Model The level tracking model in NOTRUMP consists of several features. The stratified fluid node arrangement is described in Section 2-1-1 of Reference 1-1, and the calculation of vapor flow from the p
two-phase mixture to the steam bubble using the drift flux model is described in Appendix H of l
Reference 1-1. Additional modifications made to increase code robustness are described in h
Sections 2.8,2.9, and 2.18 of this report.
1 l
1.8.1 Model Description The basic features of the level tracking model are illustrated in Figure 1.8-1. Typically, the fluid in a component is represented by a stack of fluid nodes, where the top node is either closed off or is connected to a flowlink. Figures 1.8-2 and 1.8-3 illustrate how the level tracking model is applied to two components of the AP600. Assume that the initial condition of this stack is a homogeneous two-phase mixture. At the top node, a vapor bubble (called the vapor region) is calculated to begin and
. grows as vapor rises through the mixture (when a vapor region forms in a node it is said to be stratified). The vapor-mixture interface is assumed to recede or advance at the local velocity of the liquid (U)in the mixture. 'Ihe mass flow rate of vapor from the two-phase mixture region into the i
vapor region is given by:
W, = a p,A (U,- U,),
(1.8-1) u y
1 where conditions of the mixture region (subscript M) are taken from the average conditions of the mixture region adjacent to the vapor region.
i The relative velocity (U, = U, - U,) is obtained from the mixture void fraction and tlie drift velocity
<V '> using the same constitutive models as in the drift flux models (Section 1.7). The average drift g
velocity is related to the relative velocity by:
<V > = (1 - a )U,,
(1.8-2) g u
This form of the drift velocity can be related to the weighted average <<V >> (Section 1.7, g
Equation 1.719):
l
<V > = <<V >> + (C,- 1)j, (1.8-3a) g g
= <<V >> + (C,- 1)(<xU,+ (1 - a)U,),
(1.8-3b) g Combining Equations 1.8-1 and 1.8-3b, and assuming that the liquid mixture velocity is small relative to the vapor velocity, leads to:
November 1997 c:U934wu934w.lb.NON:1b-121797
],g-1 Rev.3
- _. -- -.~_- - ~
~ _ -
l t
j
-a.e (1.8-4)
L j
Coding details are provided in Section 2.9 of this report. The form of C, and Vg>> used depends l-on the component being modeled. If the component is a venical or horizontal pipe, these quantities are based on the modified EPRI correlation.. If the component is a vessel or tank, they are based on the Yeh correlation. The table below describes how the RCS primary components are divided u
i
-(components not listed are not expected to have significant periods of two-phase flow, or are assumed to have homogeneous conditions; in particular, the ADS lines are assumed homogeneous):
PIPE (EPRI model)
VESSEL (Yeh model)
Balance Line Hot Legs _
PRHR inlet and outlet lines Cold Legs Steam generator tubes Core
~ Upper plenum, lower plenum Upper head Downcomer Pressurizer Under certain conditions, liquid may form in the upper vapor region. This liquid may be injected from flowlinks that are connected to the vapor region, or the liquid may be created as a result of steam
?
condensation. Normaily, liquid formed in the vapor region initially resides in the vapor region, then falls as droplets into the lower mixture region. When specified by input, however, special logic is employed (Section 2.7) to allow liquid flowing into the top of a stratified fluid node to fall directly -
into the mixture region. The two possible modes of liquid injection are illustrated in Figure 1.8,-2. In all connecting branches, the special logic is employed.
As the mixture level changes, it may fall below or rise above the location of a flowlink. If the interface is within a certain user-specified distance of the flowlink, the upstream condition that the flowlink " sees" transitions from that of one region to the other. The smaller the specified transition region, the higher the level must be in the main pipe before liquid enters the connecting pipe.
~ The mass interchange between the two regions of a fluid node, coupled with the mass flows into each region from flowlinks, establish the region masses and energies at each point in time. These
- conditions are sufficient to allow the determination of the region volumes at any time, given the assumption of a common pressure in the fluid node.
r The mixture level rises or falls depending on system conditions. If the system is depressurizing, liquid
~in the mixture region evaporates, increasing the vapor fraction and increasing the size of the mixture November 1997 i
0:0934wu934w 1b.NON;lb-121797 1.8-2 Rev.3
region. At some point, however, the higher vapor fraction results in higher vapor flow to the vapor region from the mixture, and the mixture level may fall.
Within a stack, only one active mixture level is allowed to form. In fluid nodes below the stratified node, special logic is employed to prevent phase separation (Section 2.18 and Appendix N of Reference 1-1) until the mixture level reaches these nodes. The fluid nodes below the node with the mixture level are assumed to remain homogeneous. However, drift flux models are applied in each flowlink between nodes so that movement of vapor through the mixture is calculated.
In the AP600, there are several flow paths connected to RCS components that represent important steam venting or heat removal paths (Figures 1.8-2 and 1.8-3 show two examples). These flow paths are the CMT balance lines connected to the top of the cold legs, the ADS 1-3 lines connected to the top of the pressurizer, the pressurizer surge line connected to the hot leg, the ADS-4 line connected to the top of the hot leg, and the PRHR line connected to the top of the hot leg. Each of these flow paths draw flow from the RCS component during the transient. Depending on the fluid conditions in the component, the flow may be drawn from a vapor bubble above a stratified mixture, or there may be significant liquid content in the flow.
1.8.2 Assumptions and Range of Applicability There are three key assumptions that are made in the application of the mixture level tracking model:
1)
Entrainment of liquid frorn the mixture region to the vapor region is assumed to be zero.
Regardless of conditions, only vapor is calculated to pass from the mixture region to the vapor region.
2)
If two-phase conditions exist in a fluid node, conditions for a stratified flow regime are assumed to exist, and a vapor region forms.
3)
In flowlinks connected to the top of fluid nodes to form a tee junction (for example, the CMT line connection to the cold leg), the upstream condition that the flowlink sees is the vapor bubble that forms in the stratified lower node (see Figure 1.8-2), unless the mixture level is within a distance D/2 of the branch connection, where D,is the branch line diameter.
Under some conditions, these assumptions may become invalid. In the following sections, an L
assessment is performed of the range over which these assumptions can be expected to remain valid, and the expected impact on results when the models are applied outside their applicable range.
a) Entrainment Above the Mixture Level If the vapor flow through a mixture is suddenly increased, the mixture level rises, and the liquid entrair;ed in the vapor increases. Ishii (Reference 1-26) characterized this entrainment (entrainment is i
November 1997 l
o:\\3934w\\3934w.lb.NON:Ib-121797
],g.3 Rev.3
_- _. -.. - ~. -. - _ -
.. -.~.
... - _ ~ -.
' defined as the flow of liquid divided by the flow of vapor above the mixture level) as occurring in three regions: a near-surface region just above the mixture level; a momentum-controlled region where larger drops begin to fall back, and all drops decelerate due to gravity; and a deposition-controlled region where the smallest drops are carried up by drag along with the vapor. A range of
~
. ' ata was successfully correlated in the momentum-controlled region, as shown in Figure 1.8-4. In this d
figure, the quantities along the x-axis are defined as:
(1.8-5a) r h
h *= I (1.8-5b).
N,, =
(1.8-5c) ia D*=E (1.8-5d)
L i
A r
i l/4 Apg (1.8-Se)
K=
v 2
, Pv (1.8-5f) l.
r sin U
l.'
<Apg,
-A=
where j, is the vapor volumetric flux in the vessel (ft/sec.), h is the distance from the mixture level (ft.), and D is the vessel diameter (ft.). It can be observed that for values of the x variable less than1 4
about.7 X 10, the entrainment is a small fraction (<1 percent) of the vapor flow, and can therefore be d
considered negligible. However, the entrainment increases rapidly above 7 X 10. Note that this is
. probably also evidence of a change in flow regime, where a well-defined mixture level no longer l
exists.
l' The boundary between a well-defined mixture level with negligible entrainment and an ill-defined mixture level with significant entrainment, based on the Ishii model, is shown in Figure 1.8-5 as a function of the nondimensional variables defined above (note: the database extends only to vessel diameters of 1 ft., therefore use of larger diameters in the correlation is likely to be misleading; a maximum value of 1 ft. is used). For a given height above the mixture level, a critical gas flow exists, above which the mixture level is not well-defined and there is high entrainment. Also shown in the figures are values of the same nondimensional variables calculated using NOTRUMP quantities for
- the AP600 2-in. cold leg break in the pressurizer. It can be seen that vapor flows are not sufficient to Jcause conditions to enter the high entrainment region. The NOTRUMP model is therefore expected to be applicable.
November 1997 o:U934w\\3934w.lb.NON:lb-121797 1,8-4 Rev.3
i b) Phase Separation at a Tee: Top Connection The CMT balance line, PRHR line, and ADS-4 line are all attached to the top of a hot leg or a cold
. leg. The tee model used in NOTRUMP to simulate these branches is subject to assumptions 2 and 3.
These assumptions may be invalid when conditions in the main pipe are such that stratified flow is not likely, or when conditions near the branch line are such that entrainment is likely.
Tests by Schrock (Reference 1-27) examined the processes of vapor pull through and liquid entrainment through a branch line connected to a larger pipe in which there is a stratified layer of
. liquid. Several branch line orientations were examined: top, side, and bottom. These tests resulted in the following correlation for a top connection (see Figure 1.8-6)-
i e
i14 r
- Sa p,
h (1.8-6)
F,**
= 0.395 bp, D,,
g where F,,, is the vapor Froude number at the branch line connection, h, is the distance to the liquid interface from the branch line connection, and D, is the branch line diameter.. The Froude number is based on the vapor flux and diameter of the branch line:
o i
W/(A,p,)
F,,,
=
(1.8-7)
/Dy i
In the main pipe, the stratified to intermittent boundary can be determined using a model by Dukler 1
l (Reference 1-24):
l r
i l/2 r
3 1
Pv
_h 3
a (1.8-8) p""
Ap,
j g
4 d(1 - a)/d(h/D)
D where in this case the Froude number is based on conditions in the main pipe. Note that the Froude i
number modified by the density ratio is equivalent to j' used in Section 1.7. That is:
f r
,m L
' Jr., = F,,, _p, (1.8-9) l.
,bp, i'
In the following equations, j* nomenclature is used.
E t
1 4
November 1997 c:U934wV934w.ib.NON:llul21797
],8 5 Rev.3 i '
- ~,
The nondimensional vapor fluxes in the branch and main lines are related by:
,. sw
.1 D,
(1.8-10)
J,.
" Jr.b y
. where N = 1 when the vapor flowing into the branch is flowing from one direction in the main pipe, and N = 2 when vapor is flowing from both directions in the main pipe. A third condition may exist, where the vapor flow in the main pipe is greater than the vapor flow through the branch. In this case, N is some number less than one.
Using the above relationships, the criteria for liquid entrainment into the branch line due to either local velocities in the branch line, or nonstratified flow regimes in the main line, can be examined on a common basis:
r j,* = 1-(1.8-11a) 4 d(1 - a)/d(h/D)
D
,m
)
h
'0.395' D
i Jr " 1-D N-) D, b
L J
r g
Figure 1.8-7 compares the' criteria for entrainment (the relationship between a and h, is obtained from the circular pipe geometry). The criteria are shown for a diameter ratio D,/D = 0.3, which is typical
- of the AP600 branch connections, but reprewnts an extrapolation of the branch line entrainment
~
. correlation (the largest diameter ratio tested is 0.06). Figure 1.8-8 shows the effect of the extrapolation. Except for the case where vapor is being pulled into the branch line from both directions in the main pipe (N=2), the vapor flow causes a departure from the stratified flow regime in the main pipe before it is sufficient to cause entrainment into the branch pipe from a stratified liquid layer.
The facility from which the data evaluated above are obtained is small scale, with a branch-to-main-line diameter ratio smaller than the AP600. In addition, the fluid velocity into the branch line is significantly higher than that expected in the AP600 because the focus of the test is on critical flow rates in the branch line.
Additional data exist from larger-scale tests, performed by Mudde (Reference 1-28). The branch line
~ diameter was 4 in., and the main line was 9 in 'in diameter, for a diameter ratio of 0.4, larger than that
' in the AP600. ' These tests were run in a different manner than the previous tests. The vapor and liquid were initially mixed in the main line, then allowed to separate (if conditions permitted) as the branch line was approached. Figure 1.8-9 shows the fraction of vapor extracted into the branch line versus the fraction of liquid extracted into the branch line for three tests in several flow regimes. It was observed that if the flow regime is stratified (test series 5), nearly all the vapor in the main line is November 1997
. n:\\3934w\\3934w-lb.NON:lb 121797
},8 6 Rev.3
~.
-,. - - ~ - - - - - - -. ~.- - _.. - --
- - -.~
s bypassed into the branch line. This supports the assumption that the branch line draws vapor from the
~
main line whenever two-phase stratified conditions exist in the main line. These results also indicate that the transition region D, identified in assumption 3 should be small, so that the branch line draws from the ' mixture only when the pipe is nearly full Since the assumption has been made in i
NOTRUMP that the [
]" and since for the AP600
.this branch line diameter is nearly one-third of the main pipe diameter, the NOTRUMP model may
.' draw from the [
]" than indicated by the data.
~l 1.8.3 Conclusions 4
It is concluded that the NOTRUMP level tracking model is applicable for the purpose of calculating the AP600 transient. Although there is no specific model or correlation to calculate entrainment into branches and tees, the data indicate that as long as the main pipe flow regime is stratified, there will be little or no entrainment into the branch until the mixture level is at the entrance to the branch. As seen in Section 1.7, the conditions calculated in the horizontal pipes for the AP600 are always 4
stratified. The simple model used in NOTRUMP is therefore adequate to calculate entrainment for these conditions. It should be noted, however, that stratified conditions may not always exist in the L tests, which typically have horizontal pipe diameters much smaller than full scale.
t i
l'.,
i l
)'
a l
November 1997 M934*\\3934w lb.NON:Ib.121m 1.8-7 Rev.3
TABLE 1.81 FIGURES DEPICTING RESULTS FROM THE LEVEL TRACKING MODEL Figure No.
Title 1.8-1 Typical NOTRUMP Level Tracking and Noding Scheme 1.8 2 NOTRUMP Modeling at Cold Leg Balance Line Junction 1.8-3 NOTRUMP Modeling of Pressurizer and Surge Line 1.8-4 Entrainment as a Function of Vapor Flow j, and Distance to Mixture Level h (Reference 1-22) 1.8-5 Non-dimensional Critical Vapor Flow for Entrainmert in Pressurizer 1.8-6 Inception of Liquid Entrainment from Stratified Layer into Branch Line (Reference 1-23) 1.8-7 Comparison of Predicted Boundary Between Low and High Entrainment with Dukler Flow Regime Model 1.8-8 Effect of Branch to Main Pipe Diameter Ratio on Predicted Entrainment Boundary 1.8-9 Phase Separation Behavior at a Tee (Top figure shows flow regime in main pipe)
From Reference 1-24 1
November 1997 o:0934wu934w-lb.NON:lb-121797 1.8-8 Rev.3
i 1
Liquid from flow link or created byN
\\e condensation e
falls as drops puu e
e Liquid from flow link j
falls directly to mixture r
(user specified) 0 O
Vapor and liquid flows based on mixture momentum equation and drift flux model O
O i
i i
Figure 1.8-1 Typical NOTRUMP Level Tracking and Noding Scheme i
November 1997 o:0934wu934w-Ib.NON:Ib 121797
],8 9 Rev.3 1j
n 0
Cold Leg Balance Line o
h Cold Leg To From RCS Reactor =
=
Pump Outlet Vessel Normal Drift Flux Model
=
" Reflux" Model Returns Liquid to Two-Phase Mixture
=
l l
Continuous Contact Flow Link Area; Donor Vapor Fraction Based on mixture Level Within Area Figure 1.8-2 NOTRUMP Modeling at Cold Leg Balance Line Junction o:u934wV934w-lb.NON:lb-121797 1.8-10 v.
1
= To ADS Valves i
l O
O O
Pressurizer Surge Line n
SG Inlet (Inclined-L Hot Leg) l.
I Figure 1.8-3 NOTRUMP Modeling of Pressurizer and Surge Line November 1997
' o:\\3934w\\3934w.lb.NON:Ib-121797
],8.))
Rev.3
-1 10 i
,, iiii, CORRELATION FOR MOME 9TUM CONTROL REGION P( MPa)
~
I.72'
~
o e 3.75 STERMAN et al.
-2 e ll. Il 10
. ig,74 3
gg
~
o 0.129 KOLOKOLTSEV of 0.101 GARNER et al.
,e o 0.11 g
v 0.30 STYRIKOVICH a 2.5 et al.
so 5.0 '
AIR -WATER *e
-e -3
'T 10
=
y w
e x
a e
-4 10 9
9 a
9^
~
~
I
~
a o'
v IA A
-5 10
-4
-3
-2 10 10 10 e 41 f pg 1-0 30 0
e a o
(Ig/h ) N.s On gg Figure 1.8-4 Entrainment as a Function of Vapor Flow j, and Distance to Mixture Level h (Reference 122) v.
o:u934wu934w-lb.NON:ltel21797 1.8-12
a 0Ny ENTRAINMENT ABOVE THE MIXTURE LEVEL 2;
(symbols are NOTRUMP k, mixture level data in the pressurizer) 3 i,
g 8.0 F
5 r
w r
[
7.0 HIGH ENTRAINMENT C
E 3
?
e:
g 6.0
=
E-g 5.0 jv 5
!!L y
4.0
~
C 1,
3 o
3.0 LOW ENTRAINMENT m
2.0
!!L e
?-
1.0 4
after ADS, increasing mixture level
_=
+
i 4
+
z y
%+ 7 after ADS, decreasing mixture level 0.0 o
e y
g-0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 K
I?".-
.0007 h* N
- Du2(pjg ).i y
p
.g
_ ~
100 i
i i
i i
i i
b T/S Air-water Steam-water l
2-U P O
IB-U P a
A
/-
I 1
/
i I
/
i 4/
1 3
/
A t
e a
f k
/
~
f
~
4 1
-N I
a
/
s O
I p
w g
i i
D I
/
10 f
/
/KfK Air-water
~
D
/
dato l
~
l
/
l
/
j
~
/
I i
/
1 i
/
3 I
I I
t t
i t
i i
10 h /d b
Figure 1.8 6 Inception of Liquid Entrainment from Stratified Layer into Branch Line (Reference 123)
November 1997
)
o:\\3934w\\3934w-lb.NON:Ib-121797 1.8 14 Rev.3 u
a s-e i
BOUNDARY STRATIFIED / LOW ENTRAINMENT TO
-g a
y
[
WAW/HIGH ENTRAINMENT; Do/Dm=0.3 1
5 eo Ei a
0.9
{
,r g 3
3 1 0.8 0
2%
e S. 7 0,7 o
+0o m
. f 5.
0.6
+Oo
+
~a n
+
g, o
0.5
+
\\
+
m
=
b w
0.4
+-
+
a
+
3
+++
g 0.3 E
4 0.2 E
0.1 so 0
0 0.2 0.4 0.6 0.8 1
1.2 1.4 1.6 1.8
-2 i
3 5
E JGSTAR
~
9 g
x 5
DUKLER
+
N=2 o
N=1 A
N =.1
?@
(S7 war.)
um 4
g m
i li BOUNDARY STRATIFIED / LOW. ENTRAINMENT TO a
y
[
WAW/HIGH ENTRAINMENT; N=1 1
5 M
E Z
0.9
~
5 3
g 0.8 e
o 0.7 E
E.
0.6 31 gs, o
0.5 Y
l D
E I
0.4 1
lc h.
0.3
$y 0.2 iE g
0.1 n.
M 3
0 g.
0 0.2 0.4 0.6 0.8 1
1.2 1.4 1.6 1.8 2
2 E
3 E
s B
m JGSTAR K
p'
[
STRAT
+
Db/Dm=.3 o
Db/Dm=.05 I
wu w
l m 1.5 i
j
'14" 8.
~/.s 5
12 13.
=
~ y. -
z
_g
.ar 10 -
- 11.. _. -
l u 1.O w
=7
.8 9
'5 Mz'A CT
=
.6 3
?.
l
.5 i
E 2
5 1
4 SW -
9 SS l
a.5 O
.1
.2 superficial gas velocity (m/s)
Flow map for the inlet and-run, as obtained from observations. SS = stratifled smooth: SW = stratified wavy; B = bubbly flow: E, measurement series.
l t
.5 -
e i
^
o m
I l
Io
- 4 I
,8,,
o o
8 r
.3 e
i
=
,I e.
, e 10
=.2
=-
A,'
a p
O
/
f.1
/o e
e
=
l A
,/ D a
l 0
O
.5 1
fraction of oss extracted (-)
Decrease of phase separation behaviour with increasing liquid inlet velocity. 3, Series 5 (uts, = 0.7 m/s, cs, = 0.08 m/s); V. series 10 (utsi = 1.2 m/s.
ucs, =
u 0.19 m/s); O. series 14 (uts, = 1.5 m/s, ucs, = 0.12 m/s);
--, Seeger model.
l Figure 1.8-9 Phase Separation Behavior at a Tee (Top figure shows flow regime in main pipe) l From Reference 1-24 i
November 1997 l
o:U934wU934w-Ib.NON:Ib 121797
],8-17 Rev.3
i 1.9 Hydraulic Resistance Model The AP600 system consists of several interconnected pipes of varying diameter. These can be divided into the following components:
RCS loops CMT balance line DVI line.
Accumulator line ADS 1-3 piping ADS-4 piping i
Typically, each component consists of a constant area piping run each with several bends and valves.
The most significant sources of pressure loss are therefore primarily due to friction and form loss.
1.9.1 NOTRUMP Model Description and Applicability i
The NOTRUMP model of the AP600 uses the Martinelli-Nelson correlation for the two-phase 2
multiplier $ in Equation 1.7-3. The resistance of each segment of pipe modeled as a flowlink is 3
supplied to the code as an overall resistance coefficient K based on single-phase flow tests or handbook calculations, therefore it includes both friction and farm loss.
The Martinelli-Nelson correlation is based on low-pressure boiling data in small diameter tubes, in which the pressure drop is dominated by friction. In the NOTRUMP model, these multipliers are applied in flow paths where form losses due to orificea, branches, and tees also exist. To establish whether this model remains applicable under these conditions, more recent correlations and models were.:xamined.
Collier (Reference 1-29) reviewed available correlations for two-phase pressure drop across orifices, bends, and large diameter pipes. One of the more successful formulations was that by Chisholm (Reference 1-30)in which the two-phase multiplier 4,2 is given by:
@l = 1 + - + -'--
(1.9-1)
X X2 where X is the Martinelli-Nelson parameter (ratio of liquid to vapor pressure drop), and C is an empirical constant. The constant C varies from 2 to abc>ut 10, depending on the component involved.
l The liquid-only multiplier and the liquid multiplier are related by:
@[ = $l(1 - x)2 (1.9-2)
November 1997 c:\\3934w\\39Mw.lc.NON:Ib-121797 1,9 1 Rev.3
where x is the steam quality. The Chisholm equation is compared with the Martinelli-Nelson correlation for two values of C in Figure 1.9-1. It can be seen that the Maninelli-Nelson correlation generally overpredicts the two-phase multiplier.
i In general, a large degree of uncertainty must be accepted in the prediction of pressure drops in two-
' phase flow regardless of the models used. Typically, even the most sophisticated correlations can only predict the pressure drop to with 30 or 40 percent. For this reason, it was oxided not to incorporate one of the newer models into NOTRUMP.
1.9.2 Conclusions t
It is concluded that the model used in NOTRUMP for two-phase pressure drop is applicable, although
'it is subject to substantial uncenainty.
t I
I i
l r
f E
i November 1997 o:0934wu934w-Ic.NON:Ib-121797 1.9-2 Rev.3 l
9 l m
G
~
e 90 t
E b.
e w
r te Comparison of Martinelli Nelson with Chisholm model for Two Phase Pressure Drop i
a n2o 7.a O 1000 z
o E. g
\\.
1 e
s' T
M m
\\
g W l2.
s ng s
w s=
\\
h *.
N, HE s '.
5 m s
N o=
s 45 100
's '.
g 5 *::=
g r
8 s
m 2 e
N o 2.
\\
'e M
\\
m Martinelli Nelson O :s a
s
--- Chisholm (C=2) p gn Q
s l5 *.i
's
- - --Chisholm(C=10) w s
o me s
g DE
's o.
x-o s
oS E
10 -
'N s
11 m s
N tJ :::
s s
s
's fI
's u o s
W n.
's Oh
- r
!! o:;-
1 y
ob
- q 0.01 0.1 1
10 o<
3 6
MARTINELLU-NELSON PARAMETER W
q r[
G lC "'
C o-
<C
- W WW
.. - ~
- -...- - - -. ~ _
i 4
i 1.10 Core Makeup Tank Model 1.10.1 Core Makeup Tank Behavior Observed from Tests Both single and separate effects tests have characterized the following behavior of the CMT during a small-break LOCA:
i 1)
After the isolation valve opens, but before significant voiding occurs in the RCS, hot RCS water from the cold leg flows into the CMT while cold CMT water flows into the DVI line l
and into the vessel. The hot water collects as a stable layer at the top of the CMT, cooling slightly as heat is transferred to the cold CMT walls. The recirculation rate is relatively low due to the low gravitational driving force.
=
2)
After significant voidmg occurs m the RCS and these voids are propagated into the balance line, a bubble forms in the CMT and the CMT begins to drain. Draining may be affected by condensation of vapor in the colder CMT water.
The physical processes occurring in the CMT during recirculation are illustrated in Figure 1.10-1.
Tests indicate that three layers form: a cold layer (the original CMT water), a hot layer (most of the water entering from the RCS), and a warm layer (water next to the CMT walls that cools and sinks to the hot-cold interface).
'1.10.2 NOTRUMP Core Makeup Tank Model (Recirculation Phase)
The NOTRUMP model of the AP600 CMT is illustrated in Figure 1.10-2. It consists of four fluid volumes, each representing the indicated fraction of the total tank volume. The basis for the choice of volumes is discussed in Reference 1-31. The key volume is felt to be the top-most volume, since its' rate of heat-up and subsequent saturation would detennine the onset of draining of the CMT.
Any warm water entering the CMT mixes completely with the water in the top volume, and this warming propagates to the other volumes as time passes. In addition, a simple lumped parameter
. (single temperature) model is used for the CMT wall.
This model ignores several potentially impoitant phenomena identified in the PIRT, the main one being that it does not recognize the fact that stable (hot fluid over cold fluid) stratification is likely to occur in the CMT.
i 1.10.3 Hot Layer Model To examine the implications of this modeling of the CMT, a more accurate " layer" model was l
developed for comparison. The basic features of the model are illustrated in Figure 1.10-1. Assume l
that the hot water entering the CMT collects in a volume V, (containing both the hot and warm November 1997 oA3934w\\3934w.lc.NON:lb-121797 1.10-1 Rev.3
. layers), which displaces the volume of cold water V,. The flow rate entering the CMT is W,, and the water leaving is W, Since the volumetric flow is conserved, the following relationship holds:
W, = bW, (1.10-1)
P,s where pu and p, (lbift.') are the incoming hot fluid density and the initial CMT fluid density, respectively. A mass balance on the hot volume gives:
E=bW*
(1.10-2) dt p,
- where p,is the density of the fluid in the region V,.
An energy balance on the hot region V,, assuming constant pressure, yields:
4 dh M,J=lpW,@,3-h,P Q,+ Q,,
O.WM dt p,,
where h is the fluid enthalpy (Btu /lb.), Q, is the heat flow (Btu /sec.) from the wall to the fluid, and Q,, is the heat flow across the interface between the hot volume and the cold volume.
As the hot volume expands, initially cold wall area is exposed to the hot fluid. De heat flux from the freshly exposed area is initially high, then decays as a thermal boundary layer builds up. The total I.
heat flow to V,is therefore given by:
y Q(t),=fP,(z)q dz (1.10-4) 0 bc >
s Where P, (Z) is the wall perimeter at location Z, and the heat flux function q (Bru/secift.2) is the heat flux at location z after an exposure time equal to (z. - z)/U,,, where U,, is the rate (ftisec.) at which the hot / cold interface travels. Since the heat transfer coefficient is relatively high (500 to 700 Btu /hr/ft.2 *F based on the CMT single effects tests), the heat flux is close to that from a semi-
/
infinite solid early in time.
' 1.10.4 Model Comparisons The two models described above were coded as simple stand-alone models. In each model, the recirculation rate is calculated assuming equal pressures at the balance line entrance and the DVI line l-November 1997 c:\\3934w\\3934w-Ic.NON:Ib 121797 -
1.10-2 Rev.3
exit, and using a momentum equation consistent with the model used. For the layered model, the recirculation rate is calculated as:
2p,g[(Zoy+ Z )Pe+ ZrsPru-Z Pd rc at W, =.A (1.10-5) bK,t+ Kay 3
P, where W, is the mass flow (Ib./sec.) out of the CMT, DV denotes the DVI line, BL denotes the balance line, TC denotes the cold layer in the CMT, and TH denotes the hot layer in the CMT.
For the NOTRUMP mcdel, the recirculation rate is calculated as:
4' 2gp, Zoyp, + { Z,p,- Z,tph (1.10-6)
W,= A bK,t+ Koy P
The NOTRUMP model was tested with the original 4 cells,10 cells of equal size, and 20 cells of equal size. Figures 1.10-3 to 1.10-7 compare some important variables from the two models.
Indicated along the time axis in these figures are the times at which the CMT is observed to begin draining in the SPES tests. Figure 1.10-3 compares the fluid temperature in the top volume of the NOTRUMP model with the temperature of the hot layer in the layer model. Until about 300 seconds, the NOTRUMP model temperature is significantly lower. This may result in excessive steam condensation onto the liquid when steam enters the CMT in the larger size breaks, leading to a delay
~
in draining. For smaller breaks, the upper cell temperature reflects the predicted actual temperature reasonably. well.
Figure 1.10-4 compares the exit fluid temperature for both models. It is evident that the thermal propagation inherent in the NOTRUMP model leads to substantial differences in the exit fluid temperature for transients in which the draining phase is delayed by more than about 800 seconds for the base model. Increasing the number of cells mitigates the problem to some extent. For smaller size breaks, the warmer water entering the vessel may impact the overall transient.
Figures 1.10-5 and 1.10-6 compare the calculated fluid energy in the tank and the metal energy.
While the simple lumped parameter model results in a different metal energy, this component is a relatively small frection of the energy change due to inlet and outlet flows. As the lowest cell heats up in the NOTRUMP model, the rate of energy accumulation decreases. Figure 1.10-7 compares the CMT flows, which agree closely.
November 1997 c:0934wu934w-Ic.NON:Ib.121797
'1.10-3 Rev.3
l t
1.10.5 NOTRUMP Core Makeup Tank Model (Draining Mode) l The NOTRUMP CMT model assumes that draining begins when the top-most volume becomes l
saturated. The tests (Reference 1-31) indicate that the processes leading to bubble formation are n. ore l
complex than a simple heatup of the upper-most CMT fluid to saturation. For the high-pressure tests L
(most representative of conditions in AP600), when the balance line drains and steam enters the CMT, time delays no greater than 50 seconds were observed, with draining beginning while the CMT upper region fluid was still subcooled.
l The delay observed in the CMT tests occurs because steam is allowed to enter the CMT without an initial recirculation period. For more realistic situations where a recirculation period occurs, these delays would be expected to be even shorter. As indicated in the previous section, the NOTRUMP model would also be expected to produce a short delay, even though the volume must heat to saturation because the initial temperature of the volume is high for relatively small breaks. Longer delays could be expected for larger breaks, but the more rapid depressurization rate of these transients quickly brings the saturation temperature down to the fluid temperature to begin draining.
I 1.10.6 Conclusions This assessment indicates that the lack of a thermal stratification model could lead to high-energy fluid being injected into the RCS from the CMT for the smaller breaks, affecting the transient. Lack of other detailed models, such as a transient thermal conduction model and a detailed condensation model, are not expected to have a significant effect on results.
i 0:0934wU934w-Ic_NON:1b-121797 -
1.10-4 v.
l TABLE 1.101 FIGURES DEPICTING RESULTS FROM TIIE CORE MAKEUP TANK MODEL Figure No.
Title 1.10 1 Stratified Layer Model of the AP600 Core Makeup Tank 1.10-2 NOTRUMP Model of the AP600 Core Makeup Tank 1.10 Fluid Temperature at Top of CMT during Recirculation: Layer Model vs. 4,10, and i
20- Node NOTRUMP Model 1.10-4 Fluid Exit Temperature during Recirculation: Layer Model vs. 4,10, and 20- Node NOTRUMP Model 1.10-5 CMT Fluid Energy during Recirculation: Layer Model vs. 4,10, and 20- Node NOTRUMP Model 1.10-6 CMT Wall Energy during Recirculation: Layer Model vs. 4,10, and 20- Node NOTRUMP Model 1.10-7 CMT Flow Rate during Recirculation: Layer Model vs. 4,10, and 20- Node NOTRUMP Model f
f November 1997 e
oA3934wu934w Ic.NON:lb-121797 1.10-5 Rev.3 i
WPhi,Thi hi Wall Heat Transfer Model:
7 1
Wall Exposure Time:
j h
_x
_ (Z - Z. ) / U Z
a; r#,
g h
J he
- j; '
s
- w. 3,
\\
.- 0 l
U hc "
V,T c
c W, p c,T c
c n
Figure 1.10-1 Stratified Layer Model of the AP600 Core Makeup Tank November 1997 o:U934wu934w-Ic.NON:lb 121797 1.10-6 Rev.3
WPhi,Thi hi Wall Heat Transfer Model:
V 10%
T T
4 m
Volume 4 15%
NOTRUMP
=
Volume 3 Volumes 15%
Volume 2 Y
60 %
Volume 1 t
W, p 3,T3 c
r r
. Figure 1.10 2 NOTRUMP Model of the AP600 Core Makeup Tank November 1997 oA3934w\\3934w-Ic.NON:1b 12t?97 1.10-7 Rev.3
T~
E. -
2!
so T
fi i
CMT TOP LAYER TEMPERATURE
~
a f,.
LAYER =1; 4 ND= 2; 10 ND= 3; 20 ND=4 b
k s.
600 I
w=
u
- e f
f 500 1
o ii H*
zCE k-@
400 o
5o E.'
E 10 9
4 300
=
e.
c 3.
A x
3.
200 2
E.
1 E
{
100
?
w, y
f DEDV i 2"DV i 2"8L i
i2"CLi i
i i
l'EL 0
g 1
0 0.2 0.4 0.6 0.8 1
. 1.2 1.4 1.6 1.8 2
5 ill x
I (Thousands) yg TIME (SECONDS)..
wm
E 5
I I
1 i
CMT EXIT TEMPERATURE LAYER =1; 4 ND= 2; 10 ND= '3; 20 ND=4 h
M5 400 5-ii E
&m 350 2OH
. :e "
H E
[a
.300 E
h&
250
=.
- a h
,I 200
- r 2
lirp 150 R
&j-100 E
8.
a 50 f
D DEDV i 2"DV i 2"BL i
i2"CLi i
1"CL 0
v 5
y v
v gy 0
0.2 0.4 0.6 0.8 1
1.2 1.4 1.6 1.8 2
o 0
E T
(Thousands)
I' TIME (SECONDS)..
<g wu i
...m
g-
=
a r
s n
t CMT FLUID ENERGY t
C g
LAYER =1; 4 ND= 2; 10 ND= 3; 20 ND=4
'a o
wn if ~
- E 45 v
2g E
8.m
.n -
3 2[-
40 OHM wR
$a' 35 i
-Ns E4 gs 30
. So
~w C
25 9
E o
8 20 j
ta i
e i
15 E
C l
10 2.
.d i
y 5
Y EDv i 2"DV l 2"BL i
i2"CLi i
i l'pl 2
o o
g 0
0.2 0.4 0.6 0.8 1
1.2 1.4 1,6 1.8 2
1 57
- e (Thousands)
I Qq TIME (SECONDS)..
l L5 i-
_y...
E.
3
=
r 4-y j
E CMT WALL ENERGY--
h.
A LAYER =1; 4 ND= 2; 10 ND= 3; 20 ND=4
- s unE-6
- .:.y Zg G
L $
z o
s
-: 2 g On r
'Y R 5
~
e -'
g4&
E1 La 4
- x e
5 n
9
=
iir 3
5
~
8
?
2 EL 3
1 6
E
',[
1 T. L i2".CLi 2".BL i
e i
i i
0 2
S S
0 0.2 0.4 0.6 0.8 1
1.2 1.4 1.6 1.8 2
o b
,{
(Thousands) ag TIME (SECONDS)..
l L$
E 3
g
'i!
a
[
CMT FLOW RATE LAYER =1; 4 ND= 2: 10 ND= 3; 20 ND=4 wn 2
f
$4 80 E
.& 3 8
2i
$z 70 ME C*
E E' 11 60 EE 8.x e e
~$
50 P
E E
G g
40 t~
30 E
8.1 20 I
f 10 J
DEDV i 2"0V i 2"BL i
i2"CLi e
i e
l'rL i
2 0
j 0
0.2 0.4 0.6 0.8 1
1.2 1.4 1.6 1.8 2
s yk (Thousands).
aq TIME (SECONDS)..
L$
i i
l-l 1.11 Passive Residual Heat Exchanger Model l
1.11,1 Ilehavior Observed from Tests During a small-break LOCA, the PRHR removes some heat from the RCS. Typically, the amount is small relative to the energy removed by the break and the ADS. As the break size becomes smaller, i
however, the PRHR may have a more significant effect. Prior to significant voiding in the RCS, the primary (RCS) side of the PRHR is liquid solid, and the flow is driven mostly by the density difference between the hot fluid approaching the PRHR and the colder fluid leaving the PRHR. 'Ihe pressu.e difference between PRHR inlet (at the hot leg) and the outlet (at the steam generator outlet plenum) is small during most of the small-break transient because the pumps are tripped soon after the break. Consequently, the gravity-driven PRHR flow is relatively low. After the RCS voids, vapor l
flows in the primary side of the PRHR, and the overall volumetric flow rate may increase.
Single effects tests indicate that the physical picture in the IRWST can be qualitatively described as follows. Figure 1.11-1.. Hot water initially at approximately 500*F flows in the primary side, and the
)
IRWST water is at 90*F to 120*F. The large overall temperature difference causes boiling to occur on i
the secondary side, and a thermal plume surrounds the PRHR tubes. Single effects tests indicate that j
the secondary-side heat transfer follows a typical boiling curve, with natural convection heat transfer when the tube wall temperature is less than about 40*F superheated and nucleate boiling heat transfer at higher wall superheats. When a variety of nucleate boiling correlations are compared with the PRHR data (Reference 1-32), the heat flux for a given wall superheat is overpredicted by all the models. This is attributed to the action of the thermal plume suppressing the boiling process.
1.11.2 NOTRUMP Model Description The NOTRUMP model of the PRHR and IRWST is illustrated in Figure 1.2-1. No attempt is made to model a thermal plume or stratification in the IRWST. This is considered acceptable for two reasons:
1)
The NOTRUMP calculation is terminated at the beginning of IRWST injection, before thermal layers can play an important role in the process. Sine: there is no net flow in the IRWST prior to draining as occurs in the CMT, the propagation of higher temperatures from the upper cell to the lower cell is not a concern as it is in the CMT model.
- 2).
The thermal plume is observed to affect the value of the heat flux, but not the basic boiling process.
l November 1997 o:09MwV9Mw-ld.NON:1b-121797
},}}-}
Rev.3
1 De heat transfer models used by NOTRUMP are:
Primary side:
a). Single-phase flow.
1 Dittus Boelter (Reference 1-33) b) Two-phase flow Shah (Reference 1-34) c Secondary side:
maximum of:
L natural convection using the McAdams correlation (Reference 1-35), or nucleate boiling using j
the Dom correlation (Reference 1-36)-
- On the primary side, the Dittus Boelter correlation is applied within its range, even though fluid velocities are relatively low (about I ft/sec.). The Shah correlation is used within its claimed range of applicability.
On the secondary side, the Thom correlation is applied well outside its range of applicability since the j-correlation pressure range is 760 to 2000 psia. However, the Thom correlation, even when extrapolated, exhibits the same characteristics as other correlations that were examined in Reference 1-37, as shown'in Figure 1.11-1. There is, therefore, no better correlation to use except one'-
' developed from the data. In addition, for the conditions expected during a small-break LOCA, the heat transfer on the secondary side is not expected to be limiting. This is illustrated in Figure 1.11-2, L
which shows the temperature drop from the primary fluid through the wall to the secondary side for l
.various fluid velocities. Velocities of about I ft/sec. are typical of those occurring during the natural circulation period. Even when the secondary side heat transfer coefficient is reduced by a factor of 10, the highest resistnnce under these conditions remains on the primary side (it has the largest temperature drop). Bis also indicates that the details of the secondary side heat transfer are not important for small-break LOCA. For higher primary fluid velocities, the heat flux across the PRHR tube rapidly increases (see Figule 1.11-3), and the discrepancy between predicted and actual heat
- transfer is likely to become larger.
1.11.3 Conclusions
- It is concluded that the NOTRUMP PRHR model contains a model deficiency that needs to be monitored to assure that excess PRHR heat transfer is not calculated.
November 1997 o:0934wu9Mw.1d.NON:lb-121797 1,11 2 Rev.3
_ ~... _. - =_.
l i
I l
l TABLE 1.114 l
FIGURES DEPICTING RESULTS FROM Tile PASSIVE RESIDUAL HEAT EXCHANGER MODEL Figure No.
Title 1.11 1 Comparison of NOTRUMP Secondary-Side Heat Transfer Correlation (Thom) with Rhosenow Correlation and Test Data 1.11-2 Effect of Variations in Primary Fluid Velocity and Secondary Heat Transfer Coefficient on PRHR Temperature Distribution 1.11-3 Effect of Primary Fluid Velocity and Secondary Heat Transfer Coefficient on PRHR Heat Flux I
l November 1997 c:0934w\\3934w-Id.NON:lb-121797
},.3 Rev.3
.. -. =. -. i l l a,b.c l l l l i l l l Figure 1.11 1 Comparison of NOTRUMP Secondary Side IIcat Transfer Correlation (Thom) with Rhosenow Correlation and Test Data November 1997 o:u934wV934w-Id.NON:lb-010598 1.11-4 Rev.3 u. -
e es s r R 3 a g- ,1; TEMPERATURE DROP ACROSS PRHR TUBE I O g. EFFECT OF SECONDARY SIDE HTC g nm 2$ 500 a 1 3 P.L 5 $g 480 '8 E< e m 460 y"5* g-440 =a n 420 y Hj2 -400 )a 380 n p $ p. 360 EE 340 kI y{ 320
== g'; 300 36 280 ? 8 260 - 240 - 4= 220 - a E 200 h Y PRIMARY INNER WALL - OUTER WALL SECONDARY N E 5 = A x., O U=.5 FT/S, THOM + U= 1 FT/S. THOM,, o U-1 FT/S THOM/10 EG w'S -....s---- - - - - - - - - - - - - - - - - - - - - - - - - ' ' - - - - ' - - - - - - - - - - - - - ' - - - ~ ~ - - - - - - ' ' - - ' '
o 9 e r = l E HEAT FLUX ACROSS PRHR TUBE [ E EFFECT OF PRIMARY FLUID VELOCITY 450 IE 5 l2: % gy 400 w 3! =, 350 CHF (ZUBER) 3 300 5' rp g 250 o c. ? g 200 &Q p 150 E H 100 S 4 9 50 0 1 2 3 4 5 6 7 8 9 10 2 O o h h ?$NA$f fdf/J? ffftM/7f (/~7/.5) xk E o THOM + THOM/10 % ~c A ' ' ^ " ' - - ' ' -
l l. 1.12 Critical Flow Model l I Critical flow occurs at the break and at the ADS valves when they open. 1.12.1 Critical Flow through Automatic Depressurization System Valves The geometry of the ADS valve package is relatively simple, consisting of gate valves and squib - valves, both of which open quickly and present a single, sudden area change in the pipe. The ADS valves open while the system is two-phase, so the flow conditions are either single-phase vapor or a two-phase mixture. A recent survey of critical flow data (Reference 1-38) compared several critical flow models to available data. It was found that none of the simpler analytical models, such as the homogeneous equilibrium model (HEM), performed satisfactorily against all the data, while some of the more sophisticated space-dependent models performed somewhat better. l A more successful application might be expected, however, for a specific geometry such as the ADS valves. Since conditions are expected to be two-phase, the HEM was chosen to calculate critical flow through the ADS valves. In Section 1.7, the effect of ignoring momentum flux terms was evaluated. It was determined that fluid velocities were low enough, such that these terms had a minor impact on the pressure gradient in all cases except for the ADS 4 piping, when both valves are open. In this case, comparison with alternate calculations indicated that the predicted vapor flow was not significantly mispredicted. Section 5 compares the model to single-effects tests, and the resulting conaparisons are shown to be acceptable. 1.12.2 Break Flow Model The critical flow through the break is subject to considerably more uncertainty due to the unknown . geometry of the break. It is expected, however, that the HEM is reasonably applicable during the l most significant portions of the transient when the conditions at the break are two-phase. Since the break area is ranged in the AP600 analysis using the Moody model as required by Appendix K, the most important issue is to ensure that the model predicts the integral test data reasonably well, so as not to introduce significant compensating error into the calculation. 1.12.3 Conclusion The critical flow model used is applicable and has been verified against tests that simulate the ADS valve geometry and the flow conditions upstream of the valves. o:09Mwu9Mw Id.NON:1b.121797 3,}7,j y
1.13 Overview of Single Effects Tests Assessment His section summarizes the assessment results presented in Sections 3 to 6. The summary examines each key model and discusses how the model basis review in Section 1 and the single effects assessments in Sections 3 and 4 either confirm the model's correctness or point to model biases or deficiencies that come into play in the integral assessments and AP600 analysis. 1.13.1 Two-Phase Flow Model In Section 1.7, it is indicated that the NOTRUMP drift flux model exhibits the following biases: [ ]" The benchmark studies in Section 3.2 confirm the expected CCFL behavior. For any given vapor flow, NOTRUMP predicts a [ ]" flow for all scales at low pressure. This means that in venical pipes there is a tendency to carry water [ ]" with the vapor, while the data would indicate liquid [ ]" flow. Studies in Section 1.7 indicate that momentum flux effects are minor in all components except the ADS 4 piping when both valves a:e open. A comparison with altemate calculations indicated that the calculated flow rate was acceptable for this component. He single effects studies in Section 4 confirm that the predicted mixture level is lower than the measured level (see Figure 4.5-2). If the predicted void fraction within the mixture is lower than the data, this explains the mixture level results. 1.13.2 Level Tracking Model As pointed out in the previous section, there is a tendency for NOTRUMP to underpredict the mixture level when using the Yeh correlation. One of the implications of the deficiency in mixture level prediction is a lower level swell than that measured in the depressurization transients. This may result in less liquid reaching high points in the system, such as the ADS lines. However, a lower predicted mixture level may result in core uncevery predictions when the data indicate that none would occur. A second bias that is identified in the level tracking model is the assumed level at which liquid is entrained into vertical connecting pipes and branches. The data indicate that entrainment does not occur until the pipe is nearly full, as long as stratified conditions exist in the main pipe and vapor November 1997 c:\\3934w\\3934w-Id.NON;lt410598 1.13-1 Rev.3 L
occur until the pipe is nearly full, as long as stratified condition.s exist in the main pipe and vapor flows remain below a critical value. The continuous contact flowlinks used in NOTRUMP may cause excessive entrainment to occur because the branch pipe draws from the lower mixture until the main pipe is approximately 1/3 empty. However, if the flow regime in the test is not stratified, NOTRUMP l may predict excessive vapor flow. Finaliy, a basic assumption in the application of the level tracking model is that the flow regime is always stratified in the main pipe. Although it is shown in Section 1.8 that the model remains applicable in large vertical tanks, the assumption in horizontal pipes may affect the entrainment into vertical branches, as indicated above. 1.13.3 Hydraulic Resistance Model A comparison with more recent two-phase pressure drop correlations indicates that the NOTRUMP model may overpredict the plessure drop, particularly if the hydraulic loss is dominated by form loss. An uncertainty of about 40 percent must be accepted in the prediction of two-phase pressure drop, regardless of the model used. Some inaccuracy in the prediction of loop flows or vapor fractions required to balance the hydraulic loss in gravity-driven flows is expected. 1.13.4 Core Makeup Tank Model De results of the assessment in Section 1.10 indicate that the lack of a thermal stratification model and the coarse noding used lead to significant differences in CMT outlet fluid temperature for small-break transients where CMT draining is not initiated until after 1000 seconds. Sections 6.4 to 6.6 compare NOTRUMP model predictions with the CMT single effects tests. These comparisons (along with those in Section 1.10) confirm that the recirculation rates and draining rates are correctly predicted. Comparisons with steam injection tests indicate that the NOTRUMP model does not sucemsfully predict the condensation process and onset of bubble formation and draining. For the most representative high-pressure tests, NOTRUMP overpredicts the time of draf og by a factor of about two. Given the complex phenomena involved and the simple model used, this result is to be expected. ' However, steam injection into an initially cold CMT is expected to occur only for larger breaks. In these breaks, the system pressure continues to drop, bringing the saturation temperature closer to the NOTRUMP fluid temperature, so that the potential discrepancy in CMT draining time is reduced. 1.13.5 Passive Residual Heat Removal Model De results of the assessment in Section 1.11 indicate that if the flow velocity through the primary side of the PRHR is greater than about 1.5 ft/sec., the heat transfer from the PRHR will be overpredicted .because the secondary-side heat transfer coefficient is overpredicted. This model deficiency must ~ therefore be accounted for in the AP600 analysis. November 1997 oA3934w\\3934w.id.NON:Ib-121797 1.13-2 Rev.3
r' l 1.13.6 Critical Flow Model 'Ihe comparisons with ADS valve tests indicate that the critical flow model used predicts the data well. Since the small-break LOCA is characterized by relatively small length-to-diameter ratio geometry, similar to the orifices used in the single and integral effects tests, the break flow model is judged applicable to the prediction of break flow as well (in any event, the break flow is ranged in the AP600 l- ) analysis).' i L i + I J i i i o:u9uwu9u..ld.NON:thl'21797 1.13-3 v.
i I l 1.14 Overview of Integral Effects Tests Assessment The integral test results are discussed below in light of the model evaluations and assessments from the preceding sections. The results are discussed in terms of the following major time periods and events, from the smallest to the largest breaks: a) Natural circulation period b) End of natural circulation c) CMT draining to ADS 1-3 actuation I d) ~ ADS 1-3 actuation to ADS-4 actuation e) ADS-4 to IRWST drain . The SPES tests are described first, then the OSU tests. Reference is made to figures in Sections 7 and 8. 1.14.1 SPES Tests a) Natural circulation 1-in. cold leg break: The predicted system pressure begins to diverge at about 1000 seconds (see Figure 7.3.2-2). The most significant energy flows during this time are from the PRHR, the steam generator, and the CMT. Cold water from the CMT affects the pressure, reducing the temperature of the fluid entering the core and reducing steam generation. Figure 7.3.2-21 shows that the CMT injection flow is well predicted through 2000 seconds; however, as indicated in Figure 7.3.2-36, the predicted CMT outlet temperature begins to increase significantly at about 1000 seconds due to the lack of a thennal stratification model. This causes the core inlet temperature to begin to diverge (see Figure 7.3.2-42) and contributes to the predicted reduction in the depressurization rate relative to the data. The average core heat input through 1000 seconds is approximately 280 Btu /sec.; Figure 7.3.2-32 shows that the PRHR heat transfer rate is about half this amount. Because NOTRUMP underpredicts the PRHR heat transfer (see Figure 7.3.2-32), this also contributes to the higher predicted system pressure. The PRHR heat transfer is underpredicted because the flow rate is underpredicted (see Figure 7.3.2-30). This happens as soon as conditions become two-phase in the PRHR inlet line. November 1997 o;U934wu934w.id.NON:1b-121797 1,14 1 Rev.3
r l-The steam generator pressure is overpredicted after about 400 seconds (see Figure 7.3.2-37). Differences in heat transfer into the steam generator are considered to be minor factors relative to the - CMT and PRHR, while the predicted system pressure remains close to the measured pressure. As the predicted system pressure becomes higher, heat transfer to the steam generator increases and the steam ( generator pressure increases. Larger breaks: As the break size increases, the energy removal rate through the break becomes more important. For the 2-in. cold leg break, for example, the core inlet temperature (see Figure 7.3.1-42) shows good i-Lagreement up to about 800 seconds, at which point the predicted value becomes higher, consistent with hotter liquid flowing from the CMT (see Figure 7.3.1-36). However, the PRHR and steam generator still play a role. In the double-ended guillotine (DEG) direct vessel injection (DVI) line break, for example, the pressure is overpredicted while the break flow rate (see Figure 7.3.4-29) is well predicted, and there is no heatup of CMT outlet water predicted prior to ADS (see Figure 7.3.4-34), - but the PRHR heat flow is underpredicted (see Figure 7.3.4-32). i b)' End of natural circulation 1-in. cold leg break: The system becomes two-phase shortly after the pumps trip at about 300 seconds. There is a prolonged period of two-phase natural circulation up until about 2000 seconds, as indicated by the steam generator tube levels (see Figure 7.3.2-6). During most of this period, the steam generator absorbs energy from the primary because its pressure is higher. Condensation in the tubes prevents significant vapor accumulation and also causes oscillations in the flow. In NOTRUMP, the same l prolonged period of natural circulation is predicted. Since the predicted primary system pressure is higher, heat transfer to the steam generator continues and so does natural circulation, as evidenced by t L the delayed draining of the tubes. Natural circulation is broken when condensation in the tubes ceases at about 1500 seconds and a bubble forms in the tubes. At this point, the oscillations cease and the core void fraction begins to i increase (see Figure 7.3.2-17). The NOTRUMP-predicted core void fraction is higher because the ' inlet core temperature is higher (high CMT outlet temperature). i Larger breaks: In the larger breaks, the extended two-phase natural circulation period is not observed. For these breaks, the system pressure quickly falls below the steam generator secondary-side pressure, terminating vapor condensation. For the 2-in. cold leg break, for example, the two-phase natural circulation period lasts until about 400 seconds (see Figure 7.3.1-6), and for the DEG DVI line break, November 1997 I o:0934wu934w-Id.NON:1b-121797 },}4 2 Rev.3 m 4
until about 100 seconds. - Because the predicted PRHR heat removal rate is smaller, the steam generator-A tubes void and begin to drain earlier in the prediction (see Figure 7.3.4-6). c) CMT draining to ADS 1-3 1-in. cold leg break: -Vapor is measured in the balance lines at 2400 seconds (see Figures 7.3.2-10 and 11), after the steam generator tubes drain. Normally, the CMTs would be expected to drain rapidly at this point. However, the test shows that there is substantial liquid held up in the CMT balance line (see Figure 7.3.2-10), while the drain rate increases and the level gradually falls (see Figure 7.3.2-5). The void fraction in the cold leg may be so low that the flow regime is not stratified (because of the small diameter pipes used in SPES), so there is substantial entrainment of liquid into the balance line. The result is a loss of net gravity driving head and carryover of liquid into the CMT, to replace that which has drained. The NOTRUMP calculation also predicts that the balance line fills with vapor after the steam generator tubes have drained. However, because the steam generator tubes took longer to drain, the vapor filling process is delayed. Because of NOTRUMP's simple entrainment model, the CMT line remains full as the balance line draws from the mixture level in the cold leg, then rapidly fills with vapor when the balance line draws from the vapor region. As soon as vapor reaches the CMT, the top node heats to saturation and the CMT begins to drain. However, because not as much liquid is entrained and held up in the balance line during draining, the predicted CMT level falls below the data and the ADS 1-3 actuation occurs sooner. As the CMTs begin to inject cold water into the vessel beginning at 2500 seconds, there is an increase in the depressurization rate (see Figure 7.3.2-2) in both the prediction and the test. The break flow rate continues to be predicted well during this period (see Figure 7.3.2-29). However, since the predicted system pressure is higher than the data, this indicates a tendency to underpredict j the break flow rate given the same pressure. Larger breaks: i Soon after the steam generator tubes drain, the CMT balance line begins to drain. As before, the . simple NOTRUMP model predicts a sudden transition, while the data show a more rapid draining than in the 1-in, cold leg break (see Figure 7.3.1-10). As the breaks become larger, the observed transition becomes more abrupt, reflecting the rapid decrease in water level in the cold leg (see Figure 7.3.410). The larger breaks do not show the CMT drain delay observed in the 1-in, cold leg break. As soon as vapor is measured in the balance line, the CMTs begin to drain. In NOTRUMP, tne flow of steam into the top node of the CMT rapidly heats the water to saturation, and the CMT is also predicted to i November T997 oA3934w\\3934w.ld.NON:lb-121797 },]4-3 Rev.3
- drain soon after vapor flows into the balance line. Even with the largest break (DEG balance line), where the CMT recirculation period is short, CMT draining occurs soon after vapor reaches the CMT. This indicates that the condensation process in the CMT during draining is not an important phenomenon that needs to be modeled accurately. For all the larger breaks, NOTRUMP predicts a late CMT drain initiation time. The predicted break flow falls below the data in the 2-in. cold leg break (looking at the slope of the integral curve in Figure 7.3.1-29) as the predicted system pressure approaches the data, then agrees with the data as the predicted pressure becomes higher again. This is also observed in the 2-in. DVI line break (see Figure 7.3.3-29) supponing the premise that there is a bias towards underpredicting the break flow rate. P d) ADS 1-3 to ADS-4 1-in cold leg break: Prior to the ADS signal, the pressurizer has drained. When the valves open, the pressurizer rapidly refills (see Figure 7.3.2-3). Water is drawn from other regions of the RCS, including the downcomer. ' However, the overall level does not drop because the accumulator injection rate increases (see Figure 7.3.2-12). Later, after the accumulator has emptied at 5000 seconds, the downcomer level falls (see Figure 7.3.2-18). In the cold legs, the level falls when ADS opens, as evidenced by the balance lines rapidly draining and filling with vapor (see Figure 7.3.2-10). On the other hand, the core collapsed level increases as cold water from the downcomer is pulled in (see Figure 7.3.2-14). l In NOTRUMP, the mass increase in the pressurizer is lower when the ADS valves open at 3700 seconds (the earlier predicted opening is due to the more rapid draining and lack of CMT mass replacement through the balance line during the previous period). This could indicate that the - mixture level swell is underpredicted. However, the total miss flow rate through the ADS valves is predicted well, judging from the integral slopes (see Figure 7.3.2-27). A more likely explanation is f that the predicted mixture level is correct (at the top of the pressurizer, allowing vapor and liquid to i be ejected), but the predicted vapor fraction of the mixture is higher. This would explain both the lower pressurizer level (higher vapor fraction in the pressurizer) and the good ADS flow prediction . despite the higher system pressure (higher quality flow). 7he same increase in core level is predicted as cold downcomer water is pulled in, with a corresponding drop in the downcomer level (see Figure 7.3.2-18 after 4000 seconds). Additional discussion of pressurizer refill is provided in the response to RAI 440.610. The break flow is predicted well until the ADS 1-3 valves are predicted to open earlier than in the test (see Figure 7.3.2-29). Because the CMT is predicted to drain more rapidly, ADS-4 is also predicted to actuate early (see Figure 7.3.2-28). i i November 1997 i oA3934w\\3934w.id.NON:ltrl21797 },14 4 Rev.3 e w w m
i l Larger breaks: In the 2-in. cold leg break and all the larger breaks, ADS 1-3 is predicted to open later than the data because of the delayed CMT drain (see Figure 7.3.1-27). The level swell into the pressurizer is underpredicted for all the larger breaks as it was in the 1-in. cold leg break. A consequence of the reduced mass in the pressurizer is that more steam and less liquid is vented from the ADS 1-3 than measured in the tests. As a result, more liquid should be retained in the downcomer and cold legs. There is evidence that this is happening in all the larger breaks. For example, in the DEG balance line break, the balance line refills at about 1000 seconds (see Figure 7.3.6-10), with a subsequent refilling of the CMT (see Figure 7.3.6-4). In the DEG DVI line break, the downcomer level is substantially overpredicted (see Figure 7.3.4-19).- l e) ADS-4 to IRWST 1-in. cold leg break: ADS-4 opens at about 5600 seconds, and IRWST flow begins shortly afterwards. Because the CMT is predicted to begin draining earlier, ADS-4 and IRWST draining are also predicted to occur earlier (see Figures 7.3.2-28 and 26). The pressurizer level drops suddenly just after 6000 seconds. This is evidence that CCFL conditions l-existed in the surge line prior to this time. As IRWST water increases the core inlet subcooling and most of the vapor generated in the core begins to flow out ADS-4, the vapor flowing through the surge line is reduced, and the pressurizer begins to drain..In NOTRUMP, CCFL conditions also end when the vapor flow through ADS 1-3 falls to zero (see Figure 7.3.2-27 and 7.3.2-3). i Larger breaks: l For the larger breaks, the IRWST flow is predicted to begin later in the transient because of the preceding delays in CMT draining. i The pressurizer drains when ADS-4 opens in the larger breaks as well. The only exception is the DEG DVI line break. In this break, the steam generation rate in the core is higher (as seen in the l' measured vapor fraction, Figure 7.3.4-17), due to the higher inlet core fluid temperature (see L Figure 7.3.4-42). In NOTRUMP, this holdup is not predicted because the core inlet fluid temperature l and steam generation rate are lower. l l~ 1.14.2 Oregon State University Tests The smallest break is affected strongly by the identified deficiencies in the PRHR heat transfer and the lack of a stratified model in the CMT. Without modifying these models, the system was calculated to repressunze and predicted ADS 1-3 opening was delayed for nearly 2000 seconds relative to the test t November 1997 oA3934w\\3914w-Id.NON;lb-121797 },]4 5 Rev.3 j l I t
~ - (see Figure 8.3.2-2A). To better assess the other NOTRUMP models, changes were made in these models to remove the amount of heat indicated by the data (the initial CMT temperature was reduced and the PRHR heat transfer surface area was increased). The description below examines results from the modified model during time periods when the modified CMT and PRHR models play a less important role. L a); Natural circulation i
- 0.5-in. cold leg break:
i There is a prolonged period of natural circulation flow up to about 2000 seconds, during which the steam generator pressure remains lower than the primary pressure and allows vapor entering the tubes to condense (see Figure 8.3.2-6). Cold water from the CMT is a major contributor maintaining the system pressure low during this time period, i Larger breaks: - Natural circulation continues for about 100 seconds in the 2-in. cold leg break, at which point the steam generator tubes begin to drain. NOTRUMP predicts the system pressure well during this time peried, even though the PRHR heat trr.nsfer is uriderpredicted (as evidenced by the higher PRHR outlet temperature, Figure 8.3.1-31). Since the CMT outlet temperature does not begin to heat up until 400 seconds (see Figure 8.3.1-34), there is not the supply of higher energy water that was observed in the 0.5-in. cold leg break or the larger-break size SPES tests. i b). End of natural circulation 0.5-in. cold leg break: Based on the steam generator tube levels, natural circulation would appear to be broken at about 1200 seconds in the loop connected to the CMT (see Figure 8.3.2-8), and at 2000 seconds in the PRHR loop. This data must be questioned for several reasons. First, there is no evidence of two-phase conditions in the PRHR hot leg since PRHR flow is undisturbed until 2200 seconds (see Figure 8.3.2-30). Second, there is no evidence of a change in core flow rate, except for a momentary one as indicated by the core inlet and outlet fluid temperatures (see Figures 8.3.2-42 and 43). This difference in timing is not seen in NOTRUMP, where natural circulation is broken on both loops at about 2200 seconds. It is likely that the system pressure drops sufficiently to cause some regions at high points of the RCS to flash. This appears to be the case in one of the CMTs in the test, and both CMTs in NOTRUMP (see Figures 8.3.2-4 and 5), as well as the balance lines (see Figures 8.3.2-10 - and 11). CMT draining begins at about 1200 seconds, while the balance lines are still nearly full of i water (see Figures 8.3.2-10 and 11). November 1997 oV934w\\3934w.id.NON:Ib-121797 1,] 4-6 Rev.3 1
1 ) v . The CMT balance lines begin to drain at 2200 seconds, which also coincides with ADS 1-3 actuation. There is not the prolonged transition period as observed in the SPES test. This is to be expected because the larger diameter cold leg in OSU is more conducive to formation of a stratified flow regime at low vapor fractions, and because the ADS flow quickly reduces the level. l ' Larger breaks: i l For the larger breaks, natural circulation ends when the steam generator tubes begin to drain (for l example, at about 100 seconds for the 2-in. cold leg break). c) CMT draining to ADS 1-3 0.5-in. cold leg break: In this test, the CMTs drain because the system pressure falls sufficiently to cause some flashing at high points. CMT-1 begins to drain at 1200 seconds (see Figure 8.3.2-4). Because the cold leg is still i full of water, the balance line supplies water to the CMT and the CMT refills. In NOTRUMP, the CMT drains, but slowly. CMT-2 does not flash in the test, but does in NOTRUMP. The predicted system pressure can be seen to respond to this excess flow in Figure 8.3.2-2. The balance lines are observed to empty soon after the steam generator tubes drain (sec ll Figure 8.3.2-10 and 11). The delayed draining observed in the 1-in. cold leg break SPES test does not I l occur in NOTRUMP. This is because the larger diameter cold legs allow a stratified flow regime to establish in the cold legs, and allow vapor to flow into the balance line with little or no liquid entrainment. Since the larger pipe diameter is more representative of the AP600, these results indicate that the prolonged draining seen in the SPES test will not occur in AP600, and that the assumption of stratified flow for these breaks remains valid. Larger breaks: L In the 2-in. cold leg break, the balance line quickly fills with vapor when two-phase conditions occur in the cold leg. The CMTs begin to drain, although there is some refilling from water carried up in j the CMT. NOTRUMP's simple entrainment model draws from the mixture level for a longer period ) (see Figure 8.3.1-10), keeping the CMT full. De inadvertent ADS test is an example of the potential for excessive delays in CMT draining l' predicted by NOTRUMP (see Figure 8.3.7-4)if there is little or no recirculation. De top fluid volume in the NOTRUMP model must heat to saturation prior to bubble formation and draining, while the test indicates that the CMT begins to drain soon after steam flows in the balance line. I i November 1997 l cA3934w\\3934w-ld.NON 1b 121797 ],]4-7 Rev.3 l l.
d) ADS 1-3 to ADS-4 0.5-in. cold leg break: V When ADS 13 opens in the test, the mass in the pressurizer rapidly increases (see Figure 8.3.2-3). NOTRUMP predicts a similar mass increase. As the pressure falls, the accumulators inject (see Figures 8.3.212 and 13). The measured and predicted liquid downcomer levels drop (see Figure 8.3.2-18) as water is pulled from the downcomer through the core into the pressurizer. The l average core vapor fraction, which in both the test and the prediction is near zero, increases (see Figure 8.3.2-17). In NOTRUMP, some of the injected accumulator water flows bi to the cold leg, refilling it. De balance line then carries some of this water into the CMT (se, c
- res 8.3.2-4 and 8.3.2-5 and 8.3.2-10 and 8.3.2-11 at 2600 seconds), delaying the draining. NOlnUMP's simple entrainment model draws from the lower mixture earlier than it should.
Larger breaks: With the exception of the inadvertent ADS, the pressurizer mass increase after ADS actuation is underpredicted (see Figure 8.3.1-3). This results in a significantly lower flow rate through ADS 1-3 for these breaks (see Figure 8.3.1-27), and in general a larger mass inventory in the downcomer and { r cold leg. The larger cold leg mass results in a larger break flow, and in several cases a partial refilling of the CMT when water flows from the cold leg up the balance line. The underprediction of level swell into the pressurizer is tied to the average vapor fraction in the core and upper plenum prior to ADS. Figure 8.3.2-17 for the 0.5-in. cold leg break test and Figure 8.3.7-17 for inadvertent ADS show that when the core mixture vapor fraction is predicted well, so is the pressurizer mass. For the other tests, NOTRUMP overpredicts the core average vapor fraction at the time ADS opens. Additional discussion of pressurizer refill is provided in the response to RAI 440.610. e) ADS-4 to IRWST drain s 0.5-in. cold leg break: The same change in pressurizer water level (see Figure 8.3.2-3) is observed in the OSU test as in the SPES test when ADS-4 opens and most of the generated steam is vented through the ADS-4. Pressurizer draining is predicted to occur more rapidly in NOTRUMP. Because of the partial CMT L refill that is predicted, ADS-4 is predicted to open late. The calculated IRWST flow rate is higher because of lower predicted pressure in the downcomer and DVI line. De calculated IRWST flow is higher than the data for some time until the pressurizer is completely empty in the test (the pressurizer is predicted to drain earlier in NOTRUMP). This overprediction of flow is due primarily to the lower downcomer pressure that results from the predicted empty pressurizer. With a partially full November 1997 oA3934w\\3934w.1d.NON:lb.121797 1,14.g Rev.3
.---. =-.. ~....,. - --. pre'ssurizer, an additional hydrostatic pressure component exists in the test that is communicated to the 'downcomer, reducing IRWST flow. Larger breaks: .1 i ' In all the larger break tests, the IRWST flow was also overpredicted. This is again attributed to lack of additional hydrostatic head from a partially full pressurizer. i ) ) ) i 1 1 \\ a i I Ndvember 1997 c1MMwu9Mw-Id.NON:)b.32j797 1.14-9. Rev.3 i
1.15 Impact of Assessment on AP600 Analysis Methodology The single and integral test assessments result in the following conclusions regarding the key NOTRUMP models and their application to AP600 small-break LOCA: ) a) Natural circulation period In general, NOTRUMP predicts the system pressure to be higher than the data, with less heat transferred to the IRWST and less heat absorbed by cold CMT water. This higher pressure causes continued heat transfer to the steam generator, preventing a bubble from forming in the tubes. j b) End of natural circulation The end of natural circulation is important because it determines the time at which the flow regime in the cold legs becomes stratified and vapor begins to enter the CMT balance line. NOTRUMP generally predicts this time to be delayed for the reasons identified above. Prolonging the natural circulation period delays the onset of CMT draining, as discussed below. 'Ihe lack of a thermal stratification model in the CMT becomes a deficiency in the smaller-size breaks, in some cases delaying ADS 1-3 actuation for many hundreds of seconds and resulting in minimal agreement with data. The effect of this deficiency, however, is to always increase the core inlet fluid temperature, which then results in an overestimate of the steam generation rate. It is concluded, therefore, that this model deficiency leads to conservative results and is acceptable for use in an Appendix K type evaluation model. j c) CMT draining to ADS 1-3 actuation j With the exception of the SPES 1-in. cold leg break, the start of CMT draining is always predicted later than the test For most tests, the delay is due to the extended period of natural circulation, while for other tests, the delay is due to the simple model used for condensation. Because the liquid holdup observed its the SPES 1-in. cold leg break test is not expected to occur in the AP600, it can be concluded that CMT draining will be predicted late in the AP600 for all break sizes. d) ADS 1-3 actuation to ADS-4 actuation In nearly all the tests, the mass increase in the pressurizer following ADS 1-3 is underpredicted, l producing minimal agreement with the data. This is attributed to the sensitivity of the level swell to the initial vapor fraction of the mixture in the core and upper plenum. This leads to minimal agreement with the ADS 1-3 mass flow rate data in the OSU tests. However, the RCS model compensates by drawing more liquid through the simulated break, with the total system mass being underpredicted. l November 1997 c: 3934w\\3934w.id.NON;lt>121797 },1 $.j Rev.3
His model deficiency may lead to nonconservative results in some cases, because a large mass held up in the pressurizer during the late stages of the transient reduces the steam flow rate through the ADS 1-3 valves, requiring mom to flow out through the ADS-4 valves. e) ADS-4 to IRWST drain i The SPES test comparisons show that the initiation of IRWST is predicted well or late (because of the delays that occur in prior phases). In the OSU tests, however, the onset of IRWST injection and its j flow rate is predicted early in some cases. This deficiency is nonconservative. \\ Based on these comparisons, the following conclusions are drawn concerning some of NOTRUMP's key models: l a) Two-phase flow and level tracking model ne two-phase flow and level tracking models predict a lower mass in the pressurizer when ADS 1-3 actuates, which results in a higher mass inventory in the downcomer and cold legs. This excess mass, i however, is ejected from the system through the break. The model therefore compensates to retain a c total mass dictated primarily by the assumed slip ratios in the core and other components. b - The prediction of excess mass in the downcomer has a conservative component in that some refilling of the CMT occurs, causing a delay of the ADS. However, as previously noted, the mass underprediction in the pressurizer late in the transient may be nonconservative. b) CMT model ne lack of a detailed model of the CMT contributes primarily to a slower than expected depressurization rate, which again causes delays in predicted ADS actuation times, and therefore a lower predicted mass inventory. i c) PRHR model The identified deficiencies in the PRHR heat transfer model are judged to not be significant, provided ' that the PRHR primary fluid velocity is maintained at less than about 1 ft/sec. In all the integral tests, this low flow rate is predicted and a conservatively low PRHR heat transfer rate is calculated. l 1 November 1997 a:u934w\\3934w-Id.NON:llel21797 1.15-2 Rev.3 l
L i [- 1.16 Key Features of the AP600 Analysis Methodology The application of NOTRUMP as an Appendix K evaluation model to the AP600 must take into . account the findings from the assessment described in this report. First, the following conservative - features required by Appendix K are applied: i a) Appendix K requirements - [ ANS 71 + 20 percent core decay heat Application of this model increases the core power by nearly 25 percent compared with the more recent 1979 ANS standard for decay heat, during the time period of most interest near the point of ADS-4 actuation. Since the mass inventory and system pressure is in large part ' determined by the steam generation rate, application of this model leads to significant conservatism. l Moody critical flow model at the break' and break spectmm Application of the Moody model results in overprediction of the break flow by about 20 percent relative to data. The integral effects tests confirm that larger breaks tend to reduce system mass to a greater extent than smaller. breaks. This model therefore is also considered to include additional conservatism, when combined with the required analysis of a spectrum of breaks. b) Additional Conservatisms to Account for Plant Geometry Uncertainties The resistances in the DVI, IRWST, CMT, and accumulator lines are set at design upper bound values to reduce the flow rate from the passive components into the RCS. In addition, the mhimum effective l critical flow area is used in the ADS critical flow calculation, and maximum resistances are used in the ADS flow paths. l Minimum containment pressure (14.7 psia) is assumed. r l c) Additional Confirmatory Checks and Assumptions to Account for Model Deficiencies The flow velocity through the PRHR primary will be confirmed to be less than 1.5 ft/sec in all AP600 simulations. In addition, the PRHR is removed from the model after ADS 1-3 actuation to further reduce the depressurization rate. If the flow through the PRHR is higher than 1.5 ft/sec. for any significant period of time, the calculation for the limiting case (minimum mass or highest PCT)is repeated with the PRHR heat transfer surface area reduced by 50 percent to account for the potential overprediction of heat transfer, i. November 1997 OA3934w\\3934w Id.NON:lb 121797 ],16 1 Rev.3
The IRWST flow will be delayed to account for potential nonconservatism i:4 the prediction of system pressure after ADS-4 actuation.. This will be accomplished by reducing the IRWST level by 6 feet. The basis for this value is described in the response to RAI 440.721(g). In summary,' the differences between predicted and actual integral test results can be attributed to one or more of the identified model deficiencies discussed in this report.' Table 1.16-1 summarizes the highly ranked component phenomena from the PIRT (Table 1.3-1) and the results of the assessment in terms of the criteria established in Section 1.5. For those areas where the agreement was found to be minimal, specific steps have been taken to address the deficiency in the AP600 analysis. November 1997 c:\\3934w\\3934w-Id.NON:1b 121797 - },j6 2 Rev.3
O f TABLE 1.16-1 h ASSESSMENT
SUMMARY
f g Component Phenomenon Assessment Results AP600 Model~ng Approach Comments o 2 ADS 1-3: Z 5 Critical flow Reasonable Minimum critical flow areas used. Poor agreement in OSU tests l due to pressurizer refill. Two Phase Pressure drop Reasonable No additional conservatism ADSl-3 in critical flow most of required. the time. Valve loss coefficients Reasonable Upper bound loss coefficients used. t i ADS 4: C Critical flow Reasonable Minimum critical flow areas used. ? Two phase pressure drop Minimal Delay iRWST drain. Possible reason for poor prediction of ADSI-3 flow, pressurizer liquid holdup in OSU tests. BREAK: Critical flow Reasonable Break size, location is ranged. ACCUMULATORS: Injection flow Excellent Upper bound loss coefficients. COLD LEGS: 2o n Phase separation at tees Reasonable No additional conservatism Pmdicted excess entrainment { required. delays CMT drain, ADS. EG A
g TABLE 1.16-1 g ASSESSMENT
SUMMARY
(Cont.) h Component Phenomenon - Assessment Results AP600 Modeling Approach Comments i VESSEIJCORE: h Decay heat N/A 1971 +20% ANS used. 5 Natural circulation flow Reasonable No additional conservatism required. Mixture level Reasonable No additional conservatism required. Mixture level underpredicted by Yeh correlation.- CMT: Circulation Excellent No additional conservatism required. Hermal Stratification Minimal Conservative Lack of model increases CMT exit { temperature, reduces core inlet p subcooling. A Draining Reasonable No additional conservatism required. DOWNCOMER: Level Reasonable No additional conservatism required. In cases where agreement is minimal, increased downcomer level delays CMT drain. HOT LEGS: Stratification, phase separation at Minimal Delay IRWST drain. Stratified model may overpredict tecs vapor out ADS 4, reduce pressurizer CCFL. '2 Eo 5 A
- = ~
OG L'S t b ~., _
i l i O l g l Y h-TABLE I.16-1 I T ASSESSMENT
SUMMARY
(Cont.) - ? ~ Component Phenomenon Assessment AP600 Modeling Approach Comments t 8 Results i F IRWST: G i j Gravity draining Reasonable U:e upper bound line resistance. f PRESSURIZER AND SURGE LINE: CCFL Minimal No additional conservatism required. Poor agreement in PRZR drain due to low vapor f flow through surge line. f Entrainment Reasonable No additional conservatism required. Level Swell Minimal Delay IRWST drain. Poor agreement is due to excess vapor content predicted in vessel. P STEAM GENERATOR: ~i r [ Natural circulation Reasonable No additional conservatism required. Heat transfer Minimal No additional conservatism required. Poor agreement in some tests due to underprediction in PRHR heat transfer, excess energy from CMT. 7 Tube draining Reasonable No additional conservatism required. PRHR: Heat transfer Minimal Check primary flow, reduce surface area if Heat transfer not overpredicted as long as primary l necessary. side is limiting. Recirculation flow Minimal during No additional conservatism required. Under predicted flow reduces PRHR heat transfer. [ j two phase flow. 2 UPPER HEAD / UPPER PLENUM. O I N Mixture level Reasonable No additional conservatism required.' 5 R
- eEG
S i w
.~ N l l* Figures 8.3.4 1 through 8.3.4-18 contain proprietary information and have been removed from l this document.- [.
- Figure 8.3-19 is not applicable.
!:j. L Figures 8.3.4-20 through 8.3.4-45 contain proprietary information and have been removed from f.: this document. P t i I 1
i OSU COMPARISONS IRWST injection Time une-1 i l 1,600 sets o 1,400 sets + o 1,200 o ss12 O M s0 1,000 1l3 saae 803 [] + j ~ ss14 m 600 + E i W 400 t i ',000 1,200 1,400 1,600 0 0 200 400 600 800 1 Predicted (sec) d 4 i i 1 Figure 8,4-5 Comparison of IRWST-1 Injection Time November 1997 c:\\3934w.8x.wpf:Ib.120397 Rev.3
OSU COMPARISONS IRWST Injection Time Utn 2 1,400 seis O g O sB13 0 O sB12 7 1,000 + M W 8 s810 0 800 sBM i y E + 600 sB14 m 4 GE 400 200 0 0 200 400 600 8.00 1,000 1,200 1,400 Predicted (sec) l l Figure 8.4-6 Comparison of IRWST-2 Injection Time November 1997 c:0934w 8x.wpf:llul20397 Rev.3
OSU COMPARISONS Average CMT Recirc Flow CMT-2 0.6 sete Q sB13 0.5 o g 0 g + seio a D 0.4 seoe 3 + C SB14 0.3 y + b 3 40 0.2 40G E 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 Predicted (Ibm /sec) 8812 Not hcluded (nows meseded het meeswomwd range) Figure 8.4-9 Comparison of Average CMT-2 Recirculation Flow November 1997 Rev.3 c:0934w-8x.wpf;1t>120397 m tt
l l I OSU COMPARISONS Average ADS 1-3 Flow 6 SBis o SB13 5 - O SB12 g W e SB10 $4 O E .o + SBos O 'b 3 SB14 Y 2 o +
- s W 2 ee3 0
1 W 0 O 1 2 3 4 5 6 Predicted (Ibm /sec) l Figure 8.4-10 Comparison of Average ADS 1-3 Flow November 1997 Rev.3 oM934w-8x.wpf.lb-120397
l 1 1 OSU COMPARISONS Average ADS-4 Flow 2 seis a 1.8 seis g'1.6 4 SB12 O W OM* SB10 E,2 o o M "1 + SB14 3 0.8 + n 8 0.6 M eE 0.4 o 0.2 .'6 0 0 0.2 0.4 0 0.8 1 1.2 1.4 1.6 1.8 2 Predicted (Ibm /sec) i l Figure 8.4-11 Comparison of Average ADS-4 Flow November 1997 c:\\3934w-8n.wpf:lb-120397 Rev.3
l OSU COMPARISONS Average IRWST Flow l l Line - 1 1.8 sets o I 1.6 - seis o _ 1.4 ss12 o $ 1.2 y e O seio l j 0 0 seos 1 l y l 3 o + + ss14 l I e 0.8 w +
- s 40 0.6 e
e30.4 0.2 0 O 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Predicted (Ibm /sec) SB13, SBoe, SB14 (Test only). Fbws not fuey developed yet Mgure 8.412 Comparison of Average IRWST-1 Mow November 1997 i-Rev.3 o:0934w-8a.wpf:lb-120397
1 l' I (9) i L i l L i i a = 1w. am [ 1 1 i l
- i. -
l i l i Figure 8.415 Comparison of System Mass for Test SB10 November 1997 Rev. 3 ' oA3934w-8x.wpf:ltrl20397 l
_-..... ~. _...... _ _ _ _. _. f, a e t 1 4 + Y (..b.c) s 1 n i 1
- 1. -
i. t i J l h a t to w w. cc i 8e .O b A i 1 i 'l Figure 8.416 Comparison of System Mass for Test SB12 November 1997 Rev.3 o:\\)934w ta.wpf:Ibl2O397
I TABLE A-3 (Cont.) NOTRUMP RAI RESPONSES RAI # Description of item Reference Where Answered RAI 440.639 There is a persistent problem in both Appendix A SPES and OSU NOTRUMP calculations caused by water entering the cold legs when there is no water there in the tests. Why does this happen? Can you present comparison figures showing the cold leg levels in the test and the calculation? Many of the NOTRUMP inaccuracies are blamed on problems with the cold leg levels. P; ease provide more documentation on what the cold leg levels are. Please evaluate how these level problems can be climinated. RAI 440.640 On page 8.3.5-4 Westinghouse Appendix A discusses Figures 8.3.5-10 and i1 and Subsection 8.3.5 points out that the calculated behavior of the balance line level is wrong beyond 1500 s. Yet Westinghouse states that NOTRUMP predicts well the PIRT items related to the balance line. This is confusing and requires more explanation. Please reassess this judgement and revise the section to clearly explain that the refill is caused by nonphysical refill of the cold legs. RAI 440.641 On paEe 8.3.6-2 in the second Appendix A paragraph, the calculated behavior of Subsections 8.3.6 and 7.3.6 CMTl is very different from the test data. Why? What does this say about NOTRUMP7 RAI 440.642 In the last paragraph on page 8.3.6-3, Appendix A 55 seconds is wrong. Perhaps it Subsection 8.3.6 should be 5.5 seconds. NovembeItev. 3 1997 cA3934m4pp twpf;lb 120197 A-37
TABLE A 3 (Cont.) NOTRUMP RAI RESPONSES Reference Where Answered Description of item RAI # 'Ihe comments contained in RAls Appendix A RAI 440.643 440.613 through 440.642 are examples Section 1 of areas in the final V&V where discussions do not appear to be adequate. Please re-examine the analyses presented in the final V&V report for areas which need additional discussion or explanation. For each case where the quench model Appendix A was used, a mixture level plot should Subsections 4.3.4 and 4.4 RAI 440.644 be included that shows the calculation results with and without the quench model. Please specify the guidelines 1 that should be followed in order to i decide whether the quench model needs to be used. Clearly state that, because of these guidelines, no AP600 analysis would ever use the quench model. RAI 440.721(a) (Issue Provide additional explanation and Appendix A raised during 7/29-significance of the consistent delay in 7/30 ACRS Meeting) NOTRUMP predicted commencement of CMT draining when compared to testing data. What is the significance of NOTRUMP% lack of a flashing model in failing to accurately predict start of CMT drain-down. RAI 440.721(b) (Issue Provide additional justification why Appendix A raised during 7/29-adverse effects from noncondensible 7/30 ACRS Meeting) gases are not a concem for AP600 NOTRUMP small break LOCA calculations. Explain where the noncondensible gases end up and why assumptions made for NOTRUMP ca':ulations are conservative. " *" k v 3 A-38 os934%=pt:ib itis 97
I l 7, I TABLE A 3 (Cont.) NOTRUMP RAI RESPONSES RAI # Description ofitesa Reference Where Answered RAI 440.721(c) (Issue Provide a thorough explanation Appendix A raised during 7/29-regarding NOTRUMP) misprediction l 700 ACRS Meeting) of mass flow out of the ADS stage I, 2, and 3 valves in the OSU experiments (and related pressurizer refill). Improve the justification as to why this deficiency is acceptable. RAI 440.721(d) (Issue Provide an explanation for Appendix A raised during 7/29 NOTRUMP) mispredictions when Section 8.3.4 7S0 ACRS Meeting) compared to the OSU test results of DVI line breaks. l RAI 440.721(c) (Issue Explain the significance and justify Appendix A l raised during 7/29-the bases for any differences in the 7S0 ACRS Meeting) Nodalization between the two integral test facilities (OSU and SPES) and the AP600. RAI 440.721(f) (Issue Provide more details on NOTRUMPi Appendix A raised during 7/29-misprediction of pressurizer drainage Section 1.16 (. 7S0 ACRS Meeting) in the OSU tests. Thoroughly explain Section 8.3.4 i the significance of this deficiency in I the code, such as non-conservatively predicting IRWST flow, and how it i. will be treated in performing AP600 calculations. RAI 440.721(g) (Issue Related to (f) above, Westinghouse is Appendix A raised during 7/29-proposing to apply a penalty in Section 8.3.4 7S0 ACRS Meeting) IRWST level. Provide a detailed explanation of how the penalty is ' determined via scaling from the OSU test data to the AP600. Justify why this is conservative. RAI 440.721(h) (Issue Provide detailed justification for not Appendix A l raised during 7/29-including momentum flux in the Section I.7.5 j 700 ACRS Meeting) NOTRUMP models. RAI 440.721(i) (Issue Provide discussion on how Appendix A raised during 7/29-NOTRUMP treats entrainment 700 ACRS Meeting) (waterspout) in branch lines. N oen o m 4= w wpcib.111s97 A-39 v l t
TABLE A-3 (Cont.) NOTRUMP RAI RESPONSES RAI# Description of item Reference Where Answered RAI 440.721(j) (Issue Justify and demonstrate why use of Appendix A raised during 7/29-Henry-Fauske/ HEM rather than 7/30 ACRS Meeting) Moody is conservative for calculating break flow through the ADS stage I, 2, and 3 valves and why this is appropriate for an appendix K type calculation. i L l
ENCLOSURE 3 WESTINGHOUSE LETTER DCP/NRCl410 AUGUST 13,1998 l-3785a duc f i}}