Inr. J. Hror Mu.\\ Tron\/ur Vol. 30. No. I, pp. 69 84. 1987 @x7-9310187 $3.00+0.

lw
Punted I” Great Bnta~n Pergamon Journals Ltd.

Heat transfer and friction in the offset strip-

fin heat exchanger
HIMANSHU M. JOSHI* and RALPH L. WEBB?
* Pennsylvania State University, Behrend College, Erie, PA 16563, U.S.A.
t Department of Mechanical Engineering, Pennsylvania State University, University Park,
PA 16802, U.S.A.

(Receitied 7 October 1985 and in final form 24 April 1986)

Abstract-This paper presents analytical models to predict the heat transfer coefficient and the friction factor
of the offset strip-fin heat exchanger surface geometry. Two flow regimes are defined-laminar and
turbulent. Based on the conditions in the wake, an equation is developed to predict transition from laminar
to turbulent flow. Flow visualization experiments were performed to identify the flow structure at transition.
The condition predicted by the transition equation corresponds to onset of oscillating velocities in the tin
wakes. Equations are developed for the Nusselt number and friction factor by writing energy and
momentum balances on a unit cell of the offset strip-fin geometry. A numerical solution is used to calculate
Nu and ,f on the fins in the laminar regime, and a semi-empirical approach is used for the turbulent regime.
Predictions are compared to data on scaled-up geometries, taken in the present study, and data on actual
heat exchangers. The models predict all data within k20”;.

INTRODUCTION wake velocities begin to oscillate. At some higher flow

THE OFFSET strip-fin is a commonly used geometry in rate, vortices are shed from the fins and are swept over
plate-and-fin heat exchangers. These exchangers are the laminar boundary layers. Thus the numerical
characterized by high heat transfer area per unit solutions are valid only for the laminar wake
condition. As reported by Shah and Webb [20], the
volume, and high heat transfer coefficients. A view of
state-of-the-art for predicting the j and f
the offset strip-fin geometry is shown in Fig. 1. Heat
transfer enhancement is obtained by periodic growth characteristics of the offset strip-fin, are the empirical
of laminar boundary layers on the fin length (1), and correlations of Wieting [ 151. These correlations
their dissipation in the fin wakes. This enhancement is consist of power-law curve fits of the j and f values for
accompanied by an increase in pressure drop because 22 heat exchanger surface geometries as functions of
Re, and the dimensionless geometric parameters l/D,,
of increased friction factor. A form drag force, due to
c( and t/D,,. These correlations have no theoretical
the finite thickness of the fins, also contributes to the
basis and should not be used outside the ranges of the
pressure drop. The geometry of the offset strip-fin is
defined by the four dimensions h, 1, s and t, or the three geometries of the surfaces used to develop the
dimensionless parameters c( = s/h, 6 = r/l and y = t/s. correlations.
It is assumed that the fins are offset uniformly, by half Two practical considerations, not addressed in
the fin spacing, as can be seen in Fig. 1. previous theoretical models, are the effects of burred
Over forty years of research has been done on the fin ends and the roughness on the top and bottom
offset strip-fin geometry. It includes data on actual walls of the offset strip-fin channel. Burred fin ends are
heat exchangers and scaled-up models [l-8], flow
visualization experiments [8-121, analytical models
[13], empirical correlations [1416], and numerical
solutions [ 17- 193. The analytical model of Kays [ 131
treats the fins as short, flat plates in laminar flow.
Comparison of the predictions of the model to data
shows only fair agreement. The numerical solutions of
refs. [17,18] are based on the assumption of zero-
thickness fins. The solution of Patankar and Prakash
[ 191 is for finite thickness fins, but its validity has not
been established by comparison with data. All three
numerical solutions assume laminar wakes
throughout the range of interest of flow rates.
However, flow visualization studies of Mochizuki and
Yagi [12] have shown that, in the range of interest, the

t To whom correspondence should be addressed. FIG. 1. Geometry of the offset strip-fin

“MT 30:t-E
69
70 H. M. JOSH] and R. L. WEBB

NOMENCLATURE

Al frontal area of unit cell Nu average overall Nusselt number,

b defined by equation (2) ‘Q’,lk
CD drag coefficient Nue average Nusselt number on the top
Dh hydraulic diameter for the offset strip- and bottom walls, h,Dh/k
tin channel, equation (7) NUP average Nusselt number on the fin
4 hydraulic diameter for a rectangular sides, h, 2s/k
channel, equation (13) AP pressure drop
f average friction factor in the offset Pr Prandtl number of fluid
strip-fin array Re Reynolds number, Re, = pvD,/p,
average friction factor on the fin sides Resh = puDi,l~, Ret = Put/P,
average friction factor on the top and Re, = pvllp, Re, = pub/p
bottom walls spacing between adjacent fins
h height of the offset strip-tin channel fin thickness
hmh, average heat transfer coefftcient on velocity based on flow area sh
the back and%ont ends of the fins velocity based on flow area (s - t)h.
average heat transfer coefftcient on
the top and bottom walls Greek symbols
ho overall average heat transfer aspect ratio s/h
coefficient i t/l
hP average heat transfer coefftcient on Y t/s
the fin sides ‘IP
shear stress on the fin sides
i Nu/Re, Prl/’ r, shear stress on the top and bottom
.le Nu,JReSh Pr113 walls
JP Nu, JRe, Pr’13 P density of fluid
k thermal conductivity of fluid p viscosity of fluid
1 tin length 1? overall surface efficiency
1,’ (V2s)lRe, Vr fin efficiency.
i

caused by the shearing process used in their The present study had three objectives:
manufacture, and the roughness results from bending
continuous metal strips into rectangular Z-shapes.
(1) To define a criterion for transition from laminar
boundary layers and laminar wakes. Flow rates
This roughness may be seen in Fig. 1. Burred fin ends
smaller than that at transition are classified as
cause an effective increase in fin thickness and
laminar and those higher are classified as
therefore in the form drag. They may also lead to
turbulent.
leading-edge separation. The top and bottom surface
roughness may cause an increase in both heat transfer
(2) To develop analytical models to predict the heat
transfer and friction characteristics in the laminar
and friction.
and turbulent flow regimes.
(3) To obtain accurate friction factor data on scaled-
up, precisely dimensioned geometries, and
conduct flow visualization studies on scaled-up
arrays to relate the flow structure in the wake to
the transition criterion described earlier.

f TRANSITION IN THE WAKES

Lc
,\--- Flow visualization studies of two collinear plates
\ [9, lo] and offin arrays [8-121 have shown that, as the
KRAN~I;
flow rate is increased, the wakes develop oscillating
velocities. These oscillating velocities affect the
transport of energy and momentum in the boundary
\ layers on the fins, although the boundary layers are
l.( LAMINAR MODEL still laminar. The change in the wake flow affects the j
and f characteristics as seen in a typical j and f vs Re,
plot, shown in Fig. 2. At small values of Re,, the wakes
Retl are laminar, and a laminar model is applicable for
FIG. 2. Typical j and f characteristics. prediction of the j and f values. As ReD increases, the j
 Heat transfer and friction in the offset strip-fin heat exchanger 71

LAMINAR
BOUNDARY LAYER

LAM INAR

FOG.3. Laminar flow on the fins and in the wakes.

and fcurves change slopes and the laminar model will any given offset strip-fin geometry was obtained from
then underpredict the data. The change of slope occurs the data of 21 heat exchangers from refs. [3-51. Their
over a small ReD range. In this study, the value of Re, geometries are shown in Table 1. The procedure is
at which the curves begin to deviate from the straight described be!ow.
laminar line is defined as the ‘point of transition’. It is The Reynolds numbers Re& corresponding to the
denoted by Ref,. The region Re, c Res is the laminar slope changes of the j and f curves are visually read
region. The character of the flow on the fins and in the from the plots of the data. These are converted to wake
wakes of the laminar condition is illustrated in Fig. 3. width based Reynolds numbers Re,* using
The region Rg > Re& is the turbulent region. This
flow condition is shown in Fig. 4. In the turbulent Re,*= Regx& (1)
region, the boundary layers on the fins are still
laminar. This was confirmed during the flow
b = t+ 1.3281/(Re1)‘,’ (2)
visualization tests that are described later. Also, plate
fin heat exchangers typically operate at 200 < ReD D = W-t)h
(3)
-c 4@00. At Re, = 4000, the value of Re, may be as h (s+h)+ht/l
high as 24,000 (for I/D, = 6), compared to
Re, = 500,000 at which transition to turbulence occurs In equation (2), b is the wake width defined as the fin
on a flat plate. Therefore, the change in the slope of the thickness plus twice the momentum thickness at the
j and f curves must be caused by changes in the wake trailing-edge of the fin. The hydraulic diameter (D,) is
flow structure and their effects on the downstream defined using the minimum cross-sectional area in the
boundary layers. A criterion to predict the changes in channel and the total heat transfer area.
the wakes will be developed. The characteristic of the flow in the wake is its
Previous studies of the wake region have not studied velocity profile. It is affected by the spacing between
the effect of the fin length (I), the fin thickness (t), and the fins (s). The width of the wake, as defined by
the tin spacing (s) on the wake flow pattern. The flow equation (2) is a function of the fin thickness (t) and the
visualization part of this study was designed to study fin length (1). To account for these effects, the Re$
these variations. values obtained from equation (1) were correlated
The correlation to predict the point of transition for with the nondimensional parameters t/l and l/s. A

NONLAMINAR LAMINAR
FREE STREAM BOUNDARY LAYER

NONLAMINAR

FIG. 4. Laminar flow on the fins and oscillating flow in the wakes.
72 H. M. JOSHI and R. L. WEBB

Table 1. Geometric parameters of 21 heat exchanger cores

Fig. Dh
No. Ref. in ref. Symbol* (mm) (s;Jr,

1 3 l&50 1.255 0.134 0.040 0.121

2 3 lo-57 : 1.562 0.135 0.036 0.100
3 3 1o-63 A 3.124 0.161 0.043 0.051
4 3 lo-58 + 1.334 0.184 0.036 0.111
5 3 l&55 X 2.230 0.244 0.032 0.067
6 3 l&67 0 1.961 0.410 0.028 0.067
7 3 1@66 T 2.531 0.461 0.023 0.051
8 3 l&68 1.681 0.471 0.048 0.106
9 3 l&70 ; 1.422 0.487 0.032 0.086
10 3 lo-56 Y 1.402 0.489 0.032 0.081
11 3 l&71 0 1.400 0.492 0.032 0.08 1
12 3 10-59 l 1.087 0.528 0.036 0.107
13 3 lC72 X 1.552 0.593 0.048 0.107
14 3 l&69 ; 1.549 0.597 0.048 0.107
15 3 l&65 1.631 0.628 0.024 0.102
16 3 l&61 8 1.295 0.670 0.040 0.084
17 4 553 (28R) 1 0.693 0.675 0.060 0.202
18 3 l&64 0 2.121 0.712 0.012 0.077
19 3 lo-62 1.176 1.0 0.020 0.041
20 5 10 s” 0.632 1.0 0.020 0.038
21 4 5-3 (20R) E 0.937 1.0 0.060 0.136

*Symbols are used in Figs. 13, 15, 16 and 17. See Appendix for definitions of Dh

power law expression was obtained, using multiple identifiable. Equation (4) correlates the observed Re,*
regression values with an r.m.s. deviation of 14%. The
relationship between Ret and the flow pattern at this
Re,* = 257(1/s)‘.23(t/1)0.58. (4) Re$ will be established later.
Table 2 shows the Ref: values read from the data, the
corresponding Ret values for thej slope change, and
ANALYTICAL MODELS
the value of Ret predicted by equation (4). The Re&
values at the j and f slope changes are equal within the Definitions
error of observation, but the j values were chosen to The analytical models to predict the j and f factors
calculate Re$ because the change in slopes of the j are developed by writing an energy and a momentum
curves were observed to be sharper and more readily balance on the unit cell shown in Fig. 5. In the offset

Table 2. Reynolds numbers at transition

Surface No. ReD at ReD at Reb at

from j slope f slope j slope Reg
Table 1 change change change equation (4)

1 1100 1000 152 154 1 .Ol

2 900 900 110 130 1.18
3 800 - 51
4 1300 800 177 148 0.84
5 900 900 95 86 0.9 1
6 800 600 97 93 0.97
7 700 600 76 76 1.00
8 700 600 115 116 1.01
9 700 600 101 116 1.16
10 1000 loo0 146 108 0.74
11 700 700 103 109 1.05
12 800 700 135 142 1.05
13 600 600 107 118 1.10
14 600 800 107 118 1.10
15 900 900 156 174 1.11
16 900 700 145 98 0.68
17 - 222
18 1100 1100 169 192 1.14
19 700 700 80 64 0.80
20 400 400 54 58 1.08
21 - 136
 Heat transfer and friction in the offset strip-tin heat exchanger 13

FIG. 5. Unit cell used in the analytical models.

strip-tin array, the velocity and temperature profiles ? is the overall surface efficiency, defined as q
become fully developed in the periodic sense, at some = 1 - [l - (&/A)]Q where A,, = surface
length downstream from the entrance. That is, they area of the fins and A = surface area of the
repeat from fin to fin. This means that the non- base plus the fins.
dimensional profiles are identical at corresponding
It is reasonable to use the approximation h,
positions on successive fins. The unit cell of Fig. 5 is
= h, = hF. The fin ends typically contain about 5 % of
chosen such that it lies in this region. In Fig. 5, h is the
the total surface area. A preliminary calculation shows
dimension perpendicular to the plane of the paper.
that a 50% error in the estimate of h, and hF would
This unit cell is an idealization of the actual geometry
result in a 5% error in h,. Rewriting equation (9) in
because it neglects the possible burrs on the fin ends
terms of tl, 6 and y, and solving for qh, gives
and also the roughness on the top and bottom of the
channel. For Fig. 5, the hydraulic diameter Dh is U+@ r?,h + 1
defined as (10)
“h0=(l+a+6) p (1,x+6/a+1)he.
4 x minimum flow area Expressing the heat transfer coefficients in
D, = (5)
heat transfer area per unit length . dimensionless form as Nusselt numbers
Thus
Nu = h,Dh/k the average overall Nusselt number
4(s - t)h/2 2(s - t)h
(6) Nu, = h,2s/k the average Nusselt number on the
Dh = (Ih + th + Is)// = (s + h) + ht/l ’
fin sides (11)
In terms of the three dimensionless parameters, D, is
Nu, = h,Di, Jk the average Nusselt number on the
D = W-Y) top and bottom walls.
(7)
h (1+x+6)’
Equation (10) becomes
The Reynolds number in the channel is defined as
1+6 D,
r/Nu = - ?rNu,
Re = PVD, (1 +c(+6) 2s
P 1
D”Nu, (12)
where u is the maximum velocity in the channel- +(l/cr+h/c(+l) Db
based on the cross-sectional area (s - t)h.
where Di, is the hydraulic diameter of a rectangular
channel of cross-section s - h, and is defined as
Heat transfer
In steady-state flow, an energy balance on the unit 4sh 2s
D;=-=_, (13)
cell gives 2(s+h) (l+a)
4(/h + th + Is)rjh, Using equations (6) and (13) in equation (12)
= 41hq,h, + 4thq,(h, + hF) + 42sh, (9) (1-Y) lf6
where
rlNu = (1+x+6) [ (1+a+6)VfNup

1
% is the average heat transfer coefficient on the
(1 +a)
tin sides (14)
+(l+l/a+d/a) Nue
h, is the average heat transfer coefficient on the
top and bottom walls Equation (14) is used to calculate the heat transfer
ha, hr are the average heat transfer coefficients on the coefficient of the offset strip-fin channel. The
back and front ends of the fin completion of the model requires a theoretical basis to
h, is the average overall heat transfer coefficient predict Nu, and Nu,. Separate models are used for the
vi is the fin efftciency laminar and turbulent flow regions.
74 H. M. JOSHI and R. L. WEBB

Friction factor values of Nu, and fr as a function of the entrance

A force-momentum balance on the unit cell of Fig. 5 length parameter (t;‘), where
is used to derive an equation for the friction factor.
1: = (1/2s)/Re, (20)
Since the unit cell is in the periodic fully developed
flow region, the net rate of change of momentum, and Re, = pu2s/p (21)
hence the net force on the unit cell is zero. So
and u is the velocity in the channel based on the cross-
4t sectional area (s - h). A curve fit of their tabled values
APA,--r,41h-z,41s-C,gpvZ yh = 0. (15)
( ! gives the following expressions.

The first term represents the force due to the Nu, = 24.2-3692(1;)-0.37 x 106(Is’)2 (22)
pressure gradient, the second and third terms are the
f, = [262-46,537(1:)+0.535x lO’(t~)*]/4Re,. (23)
shear forces on the tin sides and the top and bottom
walls, and the last term is the drag force due to the As noted previously, the present model assumes h,
finite fin thickness. = hF = hp.
A, is the frontal area of the unit cell Since equations (22) and (23) are for a = 0, it is
rp is the shear stress on the fin sides necessary to correct them for the case of a > 0. The
t, is the shear stress on the top and bottom walls following correction factors are employed
Co is the form drag coefficient.
Nu, _p NUEL~
(24)
Define the following friction factors Nu, N~ELO

&I _ fELn
f=$&$j (composite value for the array) (25)
f, .&LO'

The subscripts EL0 and ELa indicate the entrance

&=& (on fin sides) (16)
length values for rectangular channels of aspect ratios
zero and a, respectively. The corrections
(on top and bottom walls). NuELcr/NanLOand fEJfELo are plotted as functions
h=&
of ls’ with a as a parameter, in Fig. 6. The entrance
Using the above friction factor definitions in equation length solutions are taken from Shah and
(15) one obtains London [21].
Prediction of Nu, and f,. The ideal unit cell shown
AP Dh 4(21) 1 r,, by Fig. 5 has smooth top and bottom walls. The model
-x~xA,xDx- zir+&:+C,;
+pv2 4(20 h 4lh ip” rpv h assumes that Nu, and f, may be predicted using
solutions for laminar, fully developed flow in a
(17)
r~tangul~channei. That is, it is assumed that the fins
which may be solved for f to yield do not affect the boundary layers on the top and
bottom surfaces. Shah and London [21] provide these
values in tabular form and as fifth degree polynomials.
Simpler third degree polynomials were developed by
Using Al = h(s + t), and equation (7) in equation (17) the authors, assuming a constant wall temperature
boundary condition
1-Y
----(f;+a~+c,t/21). (19) Nu, = 7.45 - 16.9a +22.1a2 - 9.75a3
f = (l+a+S)(l+y) (26)

This is the final expression for the average overall f, = (23.94- 30.05a+32.37a2 - 12.08a3)/ReSh (27)
friction factor in the unit cell of Fig. 5. Now one must where
establish methods to predict fr, f, and Co, for the &?,h = ReD/B (28)
laminar and turbulent regimes. and
1+a
BG------. (2%
ltafb
The laminar flow model applies when Re, < Reg.
The component Nusselt numbers and friction factors, Thus, the RHS of equations (24) and (25) are read
and Co are predicted as explained below. from Fig. 6.
Prediction of Nu, and fp. Sparrow and Liu [18] The drag coefticient (C,). For finite fin thickness,
have numerically predicted Nu, and fp for the case of 6 > 0, the pressure difference on the up and
c( = 6 = y = 0. A zero aspect ratio (a = 0) corresponds downstream blunt faces results in a form drag. Kays
to the condition s & h. The present model uses a curve [13] used Cb = 0.88 in his approximate model. This
fit of the Sparrow and Liu values for a = 6 = y = 0, results from the solution for potential flow [22]
and develops correction factors that apply when a, 6 normal to a long, thin plate, assuming the pressure in
and y are non-zero. Sparrow and Liu present tabled the wake is equal to the free-stream value.
 Heat transfer and friction in the offset strip-tin heat exchanger 15

I.000 I I I lllll
0.6 _ ’ I I lllll
0.1 0.2 0.3 0.4 0.6 0.8 1.0 2 3 4 6 8 IO

a:x IO3
FIG. 6. Correction factors for finite aspect ratios.

Experimental data [23] for the same geometry show Thus, from equations (30) and (31) six values of j,
that Co is approximately 2.0. However, qualitative and fp are obtained for each of the 21 surfaces. These
justification can be provided for using Co of the order are now correlated using multiple regression, with the
of 0.8. Roshko [24] states that a blunt plate and a Reynolds number ReD, and the dimensionless
circular cylinder, oriented normal to the flow, are geometric parameter l/Q,, as independent variables.
members of a basic family. The two geometries differ in The following equations were obtained
their ‘degree of bluffness’. For Re, > 1000, Co = 1.0
j, = 0.36(ReD)-o.433(I/Dh)-0.174 (34)
for a single circular cylinder. However, if one
calculates the Co on one cylinder in a staggered array &, = 15.33(ReD)-o~785(~/Dh)-0~324. (35)
with pitch/diameter = 3, one finds Co z 0.4. So, Co
Prediction of Nu, and f,. These values are
for the array of cylinders is only 40% of that for a
calculated by equations (32) and (33). It is assumed
[Link]. Similarly, 40% of the Co for flow
that the fins do not affect the boundary layers on the
normal to a thin plate is approximately 0.8. Additional
top and bottom surfaces.
confirmation for use of Co = 0.8 was established in the
development of a semi-empirical correlation for the
friction factor of the offset strip-fin array [25]. Values EXPERIMENTAL PROGRAM
of 0.7 < Co < 0.9 were tried, and Co = 0.8 gave the
best correlation. The present model uses Co = 0.8 for The experimental portion of this study had two
the laminar and turbulent regimes. objectives:

(1) To take accurate friction factor data for precisely

Turbulent models dimensioned offset strip-fin geometries.
For Reb > Re,*, a semi-empirical approach is used (2) To conduct flow visualization experiments to
to calculate Nu, and jr. study the flow on the fins and in their wakes, and
Correlations for Nu, and fp. One may express observe the transition from laminar to turbulent
equation (14) in terms of the heat transfer j-factors flow.
(j = Nu/RePr1’3) and solve it for qP

(1 +cr+6)’ Re, (l+cr)cr Resh Friction factor

1+Sge. (30) A majority of the data available for the offset strip-
“jp = (1 +cc)(l -y) Re, w- s
fin were taken on industrially manufactured heat
Equation (19) solved for fp gives exchanger cores [l-6]. The possibility of burred fin
edges results in a degree of uncertainty regarding the
f = u+a+N1+Y)f_af_&t
P e (31) actual fin shape. Data available for scaled up arrays
1-Y 21 .
with precise dimensions [7,8] do not include a
For each of the surfaces of Table 1, six data points systematic variation of the geometric parameters with
are selected such that Reb > Re$, where Reg is values in the ranges of practical interest.
calculated from equation (4). The reported j and f Eight models of the offset strip-fin were built for
values are substituted in equations (30) and (31). The studying. Their dimensions are shown in Table 3. Data
values j, and f, are calculated using equations for fully were also taken for one plain tin array.
developed turbulent flow in a rectangular channel The test sections were made with brass fins and
plexiglass walls. The fins were made in the form of long
Nu e = 0 .023(Resh)0.8Pr0.4 (32) brass plates with windows. These plates were placed in
f, = 0.079(Resh)-0~25. (33) the slots made in the top and bottom walls, so that the
 H. M. JOSHI and R. L. WEBB

Table 3. Dimensions (mm) of test sections

Test-section Symbol 1 t s 6 Lx I.
Dh i

Plain 762.0 0.813 4.261 7.669 0.001 0.123 0.191

1 25.4 0.406 4.674 7.518 0.016 0.123 0.087
2 0 12.7 0.406 4.674 7.518 0.032 0.123 0.087
3 n 25.4 0.813 4.267 6.096 0.032 0.112 0.191
4 0 12.7 0.813 4.267 6.096 0.064 0.112 0.191
5 + 50.8 0.813 9.347 13.487 0.016 0.246 0.087
6 25.4 0.813 9.347 13.487 0.032 0.246 0.087
7 z 12.7 0.813 9.347 13.487 0.064 0.246 0.087
8 Y 25.4 1.626 8.534 10.897 0.064 0.224 0.191

h = 38.1 for all test sections.

Symbols shown above are used in Figs. 9, 10 and 14.

windows in one plate were aligned with fins in the temperature. The properties were curve fitted to be
adjacent plate. The sides of all windows were filed by used in data reduction. The temperature of the fluid
hand to ensure burr-free fin ends. The method of was measured at the upstream end of the entrance
assembly is shown in Fig. 7. Figure 8 shows the section. An expansion tank was used to remove air
completed test section with all the walls in place. The from the system before taking data. The pressure
test sections were placed in the flow loop shown in Fig. drops (AP,, AP,) were converted to Fanning friction
9. An entrance section is provided at the upstream end factors as follows
to allow a smooth transition from the circular piping.
Aqueous ethylene glycol (65-70x by weight) was (36)
used as the test fluid. One of two orifices measured the
flow rates, controlled by the valves shown in Fig. 9. An where L is the length over which AP is measured. The
inverted manometer, with a fluid of 1.75 specific velocity u corresponds to the minimum cross-sectional
gravity was used to measure the pressure drop across
the orifices. The pressure drops across the test section
(AP, and AP,) were measured by an inclined
manometer for heights smaller than 25 mm, and by a
vertical manometer for larger heights. Fluid of specific
gravity 0.75 was used in these manometers. The test
sections had 15 fins in the flow direction, and five fins
per row in the spanwise direction. AP, corresponds to
the first eight fins of the test section, and AP, to the last
seven fins where the flow is expected to be periodic
fully developed.
The particular mixture of ethylene glycol that is
chosen gives the desired Reynolds numbers for the 1 bottom plate
flow rates availablefrom the pump. Before running the
tests, the densities and viscosities of various FRONT VIEW
compositions were measured as functions of FIG. 8. Assembled test section.

ISOMETRIC VIEW
FIG. 7. Construction of test sections for ,f data
 Heat transfer and friction in the offset strip-fin heat exchanger 77

VALVE
TEMPERATURE MEASUREMENT POINT
AP PRESSURE DROP

EXPANSION TANK

FIG. 9. Flow loop used for f data.

area in the test section developing region (AP,) were found to be S-8 % higher
than those based on AP,. The plain fin data agree
rit (37) within 2 % of the analytical solution of Curr et al. [27].
U=(S Figures 10 and 11 show that for the same aspect ratio.
geometries with higher t/I have higher friction factors
where rigis the mass flow rate between two fin plates. because of the contribution of the drag term Co t/21. As
Results for the eight test sections and the plain fin Re, is increased, the relative contributions of fp and f,
are shown in Figs. 10 and 11. The symbols correspond [equation (19)] decrease and the form drag begins to
to those in Table 3. These friction factors are based on dominate, resulting in f being almost constant. This is
the pressure drop AP,. The friction factors in the evident for surfaces 3, 4, 7 and 8.

0201’ ’
I I I

b.10 -

0.08 -
0.07 -
0.06 -

0.05 -

- 0.04 -

0.03 -

0.02 -

0.0 I - I L
200 300

FIG. 10. f vs ReD for four offset strip-fin test sections (14 in Table 3), and one plain fin. (Symbols arc
defined in Table 3.)
78 H. M. JOSHI and R. L. WEBB

i- _ ,_....T._Ilrl-r ..._~r--.--'-‘T__
Symbols defined in
Table 3

008
0.07\-
1 x

0.06/-
0.051
/
'c 004t-

0031

0.02-

00 I i I l/II I .L
20r_---” 300 500 700 IO00 2000 3oco4( 00

%I
F‘IG. 11, j’ vs RQ for four offset strip-fin test sections (5-8 in Table 3). (Symbols are defined in Table 3.)

Results of the visualization experiments are shown

Visualization experiments were done on three in Fig. 12 and Table 5. Figure 12 shows that four
scaled-up arrays. These arrays were made from distinct flow patterns may be observed as flow rate is
plexiglass fins and a wooden base. Rectangular pieces increased. In Fig. 12(a) the wake is smooth and
of plexiglass were placed in slots cut into the base to laminar. Figure 12(b) shows that oscillations first
form an offset strip-fin array. The method of assembly occur at the upstream of the second fin. These are
was similar to that used for the friction test sections, carried to the upstream fin until, at a sufficiently high
except that the fins were made individually, and that flow rate, vortices are shed. Table 5 shows three
there is no top plate. The dimensions of the arrays are Reynolds numbers for each of the flow patterns
shown in Table 4. The top is left open to facilitate observed. These values are for the 18th row of fins.
visualization. These arrays are placed in a water
channel 30.5mm wide and 4.6 m long. The wooden
base fits in the channel leaving 3 mm gaps on either FLOW
side. The arrays were composed of 20 fins in the flow
direction and three fins per row in the spanwise
direction. Black ink was used as the dye. It was
introduced into the flow with a hypodermic needle at
(a)
the end of a long vertical plastic tube. The needle was
attached by a very thin tape to the downstream face of
the fin whose wake was to be studied, This
arrangement allowed introduction of the ink using
1
(b)
gravity feed and without causing disturbances in the
flow.

Table 4. Geometries of arrays used in flow visualization tests

_.__II_
No. I r s h t/l

1 127.0 2.54 72.50 150.25 0.02 Cd)

2 63.5 2.54 72.50 150.25 0.04
3 127.0 6.35 84.14 150.25 0.05 FIG. 12. Flow patterns observed in the vis~lization
______ -~ experiments.
 Heat transfer and friction in the offset strip-fin heat exchanger 19

Table 5. Reynolds numbers at the flow conditions of Fig. 12, The flow patterns observed in these experiments are
for the three arrays of Table 4 to be related to the transition defined by equation (4).
Array 1 Array 2 Array 3
It is expected that the laminar models begin to
Fig. t/l = 0.02 t/l = 0.04 t/r = 0.05 underpredict the data when the wake becomes
unstable. Thus, the changes of slopes of the f and j
12(a) Ret 15 15 20 curves should correspond to the flow condition of Fig.
Reb 42 35 47
12(b) which shows the onset of oscillatory flow in the
Rel 750 375 400
wakes. Application of equation (4) to the three arrays
12(b) Re, 28 25 48
gives results that are shown in Table 6. The predicted
Reb 71 58 73
Rel 1400 625 960 values of Ret are reasonably close to those observed
for the flow pattern of Fig. 12(b). Equation (4) was
12(c) Ret 58 62 107
Ret. 130 116 168
developed by visually estimating the slope change
Rel 2900 1550 2140 points of 21 surfaces. Within the accuracy of these
estimates, one may conclude that it predicts
12(d) Ret 67 68 133
R% 144 123 202 reasonably well the condition of the onset of
Rel 3350 17cKl 2660 oscillations in the wakes, and that this condition
corresponds to the transition from laminar to
turbulent flow as defined in Figs. 3 and 4.
Data taken for the second row of fins were found to be
identical.
PREDICTION OF j AND f VALUES
The three arrays of this experiment allow one to
compare the effect of varying length and thickness. For Heat transfer
arrays 1 and 2 (same t), a change in length does not Equation (14) is used to predict thej-factors for the
affect the Re, values. Reb is smaller for array 2 because 21 surfaces of Table 1. The ratio of the predicted j
of smaller wake widths as defined by equation (2). For values (jpred) to the reported j values (jeXP) with
arrays 1 and 3 (same I), both Re, and Reb are higher q = qf = 1 is plotted vs the Reynolds number (ReD) in
while Re, is lower, for the thicker plate. Fig. 13. Equations (22), (24) and (26) are used for
Reb 6 Re,* (laminar range) and equations (32) and (34)
are used for Reb > Ret (turbulent range). Except for
Table 6. Comparison of equation (4) to flow visualization
results
two surfaces the predictions are within *20x. The
r.m.s. deviation of the ratios from the [Link] is 11.5 %.
Reg Observed Ret, for
Array equation (4) Fig. 12(b) wake Friction factor
Equation (19) is used to predict the friction factors
1 43 42
2 22 35 for the eight scaled up geometries of Table 3, and for
3 59 47 the 21 heat exchangers of Table 1. For the laminar
range (Reb I Re,*), equations (23), (25) and (27) are

Table I

FIG. 13. Heat transfer predictions for the Table 1 data.

80 H. M. JOSHI and R. L. WEBB

1.4 I I I I I 1 I

1.2 - 0
d

0.6 - I I I I I I I
0 500 1000 1500 2000 2500 3000 3500 4000

Rerl
FIG. 14. Friction factor predictions for the Table 3 data.

used. For the turbulent range (Reb > Re,*), equations ratio&red/& is plotted vs the entrance length
(33) and (35) are used. The predictions are plotted as parameter 1,’ used in equation (22). At approximately
the ratio fpred/fexp vs ReD. I,’ = 0.0012, the predictions start to fall below the
Table 3 data. The predictions for the data of the data, with increasing deviation as I,’ is decreased (ReD
present study are shown in Fig. 14. All predictions are is increased). This figure clearly shows that a laminar
within f 20 % ofthe data. The r.m.s. deviation is 9.5 %. model should not be used for Re, > Re,* as defined by
Equation (4), used to predict Ret, was developed equation (4).
using the 21 heat exchanger cores of Table 1. Its ability
to predict Ret for the Table 3 geometries is examined
in Table 7. Except for surface 7 the predicted Ret PRACTICAL CONCERNS
values agree with the Reb at the observed slope Effect of burrs and roughness
changes within f20%. Comparison of Figs. 1, 5 and 7 shows that the unit
Heat exchanger data. Friction factor predictions cell and the model test sections are idealizations of the
for the 21 heat exchanger surfaces are shown in Fig. 15. actual geometry. There are two major differences: (1)
The predictions are within + 20 % of the data for 16 of the tin ends in an actual core may be burred; and (2)
the 21 surfaces. The r.m.s. deviation of all predictions there is a roughness on the top and bottom walls ofthe
is 16.8 %. Examination of Fig. 15 shows a tendency to actual surface. If these two factors significantly affect
underpredict the friction factor. This would occur if the heat and momentum transfer, it would be more
burred fin edges existed in some of the cores. evident in the turbulent than in the laminar regime. An
assessment of the burr and roughness effect on the
friction factor was made for the turbulent regime as
Laminar model follows. Burrs and roughness do not exist for the Table
If the laminar model for heat transfer is used to 3 geometries. The same method used to develop
predictj over the entire ReD range, it is expected that equation (35) was applied to the Table 3 turbulent
for Reb > Ret, the predictions will fall below the data. regime data. The resulting equation is
A plot of the j predictions using the laminar
equations (22), (24) and (26) is shown in Fig. 16. The & = 3.78Re~“.62(l/DJo.23. (38)

Table 7. Comparison of equation (4) to Table 3 data

ReD at Reb at R&

Surface slope change slope change equation (4) Rei?lReb
2cQO 211 188 0.89
1400 140 119 0.85
1300 271 313 1.15
1000 197 199 1.01
1900 221 188 0.85
1400 148 119 0.80
1500 140 76 0.54
1100 231 200 0.87
 Heat transfer and friction in the offset strip-fin heat exchanger 81

1.6

FIG. 15. Friction factor predictions for the Table 1 data.

Using equations (33) and (38) to predict Jr for the values. Hence, it appears that the combined effect of
turbulent regime of the Table 1 geometries, a plot of burrs and roughness on the friction factor may be, on
&redl&p was made (Fig. 7). One is interested in the average 414% in the turbulent regime.
determining whether there is a significant In the case of heat transfer, the roughness should
underprediction on Fig. 17, as compared to Fig. 15. A cause an enhancement, whereas burred fin ends may
comparison of corresponding points on Figs. 17 and cause leading edge separation, and a decrease in the
15 show 9-20x underprediction at the low Re, end heat transfer. These two effects tend to cancel. If only
and O-8% underprediction at the highest values of the roughness is present, the surface may show a small
Re,. There is no obvious reason why the enhancement. Typically, a 50% increase in the heat
underprediction would be greater at the lower Reo transfer coefficient on the top and bottom walls would

I I I IIIII ~ I I I III/I
~m:=o.ool2 Symbols defined in
1.8 - Table I

[Link]~IIIrIll~ I I Illllll
0. I 0.2 0.3 0.4 0.6 0.6 1.0 1.2 2 3 4 6 8 IO
a; x IO3

FIG. 16. Heat transfer predictions using the laminar model for the Table 1 data.
 tY
82 H. M. JOSHI and R. L. WEBB

1.6
-7 I I I ~_T__r---- 7-.

c
Symbols defined in
Table I. I

1.4

--

08
1
x

FIG. 17. Friction factor predictions using equation (38) for the Table 1 data.

result in a 2-5% increase of the composite heat The r.m.s. deviations are 8.7x, 8.6x, 13.1% and
transfer coefficient (h,). 8.1x, respectively. The correlations predict 80% of
Whether burrs existed on the surfaces of Table 1, the f data and 75 % of thej data within + 10% for the
and the size of such burrs is unknown [28]. This laminar regime, and 88 % of the f data and 97 % of the
represents an uncertainty in the data of these surfaces. j data are predicted within f20% for the turbulent
regime.
Empirical correlations Overall 82 % of the f data and 91% of the j data are
Empirical curve fits of the data are easier to use than correlated within f 15 ‘4. The Wieting correlations
the models described above and may be preferred by predict 85% of the data within + 15 % for both f
the industrial designer. The empirical j and f and j.
correlations of Wieting [15] were previously
mentioned. He chose ReD < 1000 for his laminar CONCLUSIONS
correlation and Re, > 2000 for the turbulent
correlation. The reduced data use the D, definition of An equation was developed to predict the
Kays and London [3] as described in the Appendix. In transition from laminar to turbulent flow in the
the present study similar correlations were developed offset strip channel. Flow visualization experi-
using the Table 1 data and the criterion for Reg ments showed that the transition occurs at the
[equation (4)] as the limit on the laminar correlations. onset of oscillating velocities in the wake.
Thus, for ReI, 5 Ret: (Ref, corresponds to Re,*), the Analytical models were developed to predict the
laminar equations are used. Because of the uncertainty heat transfer coefficient and the friction factor in
in reading the j and f plots for the 21 surfaces, the the offset strip-fin. For the laminar region the
lower limit on the turbulent equations was chosen as model is based on a numerical solution done by
Re6+ 1000. The following equations were obtained. Sparrow and Liu [ 181, and a semi-empirical
method was used for the turbulent region. The
r.m.s. deviation of the predictions of the heat
Laminar range (Re, 5 Re$,):
transfer model for data of 21 heat exchangers is
f = 8.12(Re,)-0~‘4(~/Dh)-o~41(a)-o~02 (39) 11.5 %, and that of the friction model is 16.8 %.
Friction factor data were taken on eight scaled-up,
j = 0.53(ReD)-0~50(l/Dh)-0~15(tl)-0.14. (40)
idealized geometries. The friction model predicted
Turbulent range (Re, 2 Rer*,+ 1000): these data with an r.m.s. deviation of 9.5 %.
Flow visualization experiments were done to study
f = 1.12(ReD)-0~36(l/Dh)~o~65(t,JDh)0~17 (41)
the flow in the fin wakes and its effect on transition.
j = 0.21(Re,)~0~40(l/Dh)-o~z4(t/D~)o~oz. (42) Three geometries were tested. Flow patterns in the
 Heat transfer and friction in the offset strip-fin heat exchanger 83

wake may be correlated by using a wake width 14. S. V. Manson, Correlations of heat transfer data and of
based Reynolds number. friction data for interrupted plane tins staggered in
successive rows, NACA Technical Note 2237 (December
5. Empirical correlations were developed to calculate
1950).
the heat transfer coefficient and friction factor. 15. A. R. Wieting, Empirical correlations for heat transfer
These correlations are based on a different and flow friction characteristics of rectangular offset-fin
definition for hydraulic diameter than that used in plate-fin heat exchangers, J. Heat Transfer 97, 488490
(1975).
the correlations of Wieting [15], because the area
16. H. Usami, Private communication, Fuji Heavy
of the fin ends has been included. Overall, 82 % of Industries Ltd., Tochigi, Japan (July 1983).
the f data and 91% of the j data are correlated 17. E. M. Soarrow. B. R. Balina and S. V. Patankar. Heat
within f 15%. The change from laminar to transfer-and fhtid flow analysis of interrupted wall
turbulent equations is based on the equation channels with application to heat exchangers, J. Heat
Transfer 99, 4-l 1 (1977).
developed to determine transition.
18. E. M. Sparrow and C. H. Liu, Heat-transfer, pressure-
drop, and performance relationships for in-line.
Acknowledgemenr-This work was performed under staggered, and continuous plate heat exchangers, Iut. J.
Department of Energy contract DE-AS02-79ER10394. Heat Mass Transfer 22, 1613-1625 (1979).
19. S. V. Patankar and C. Prakash, An analysis of the effect
of plate thickness on laminar flow and heat transfer in
interrupted-plate passages, Int. J. Hear Mass Tran$r
REFERENCES 24, 1801-1810 (1981).
20. R. K. Shah and R. L. Webb, Compact and enhanced
1. U. T. Joyner, Experimental investigation of entrance
heat exchangers. In Heat Exchangers Theory and
reeion heat transfer coefficients. NACA ARR 3K01
Practice (Edited by J. Taborek et al.), pp. 440-441.
(1943).
Hemisphere, Washington (1983).
2. R. H. Norris and W. A. Spofford, High performance fins
21. R. K. Shah and A. L. London, Laminar Flow Forced
for heat transfer, Trans. Am. Sot. mech. Engrs 64,
Convection in Ducts. Academic Press, New York (1978).
489-496 (1942).
22. L. M. Mime-Thomson, Theoretical Hydrodyrlamics, 4th
3. W. M. Kays and A. L. London, Compact Heat
edn, p. 319. Macmillan, New York (1960).
Exchangers, 3rd edn. McGraw-Hill, New York (1984).
23. S. F. Hoerner, Fluid Dynamic Drag, pp. 3316. Published
4. F. M. Walters, Data Item No. 63.02, AiResearch
by the author, Midland Park, NJ (1965).
Manufacturing Co., Document NO. AP-69-5348
24. A. Roshko. On the wake and drae of blufI bodies. J.
(August 1969).
aeronaut. Sci. 22, 124132 (1955). -
5. A. L. London and R. K. Shah, Offset rectangular
25. R. L. Webb and H. M. Joshi, Prediction of the friction
plate-fin surfaces-heat transfer and flow friction
factor for the offset strip-tin matrix, Proc. ASME-JSME
characteristics, J. Engng Pwr 90, 218-228 (1968).
Thermal Engineering Joint Con,ference, ASME. New
6. S. Mochizuki and Y. Yagi, Heat transfer and friction
York, Vol. 1, pp. 1801~1810 (1983).
characteristics of strip-fins, Heat Transfer-Japan. Res.
26. S. W. Churchill and R. Usagi, A general expression for
6, 36-59 (1977).
the correlation of rates of transfer and other phenomena,
7. E. M. Sparrow and A. Hajiloo, Measurement of heat
[Link].E. JI 18, 1121-1128 (1972).
transfer and pressure drop for an array of staggered
27. R. M. Curr, D. Sharma and D. G. Tatchell, Numerical
plates aligned parallel to an air flow, J. Heat Transfer
predictions of some three dimensional boundary layers
102,426432 (1980).
in ducts. In Computational Methods in Applied
8. R. S. Mullisen and R. I. Loehrke, Enhanced heat transfer
Mechanicul Engineering, Vol. 1, pp. 143-158 (1972).
in parallel plate arrays, 21st National Heat Transfer
28. R. K. Shah, Privatecommunication, Harrison Radiator,
Conference, Seattle, WA (July 1983).
Lockport, NY (October 1984).
9. D. B. Adarkar and W. M. Kays, Heat transfer in wakes,
TR No. 55, Department of Mechanical Engineering,
Stanford University (April 1963). APPENDIX
10. R. I. Loehrke, R. E. Roadman and G. W. Read, Low
Reynolds number flow in plate wakes, ASME Paper No. The j, f and ReD values in Table 1 were calculated using
76-WA/HT-30 (December 1976). & and Redefined by equations (7) and (8), respectively. This
11. R. I. Loehrke, J. C. Lane and R. L. Zelenka, Heat &, definition uses the actual heat transfer surface area (A)
transfer from interrupted plate surfaces, TR HT-~~791, (which includes the blunt fin faces), and the true minimum
Department of Mechanical Engineering, Colorado State flow area [A, = (s-t)h]. Using equation (6) for Dh, the
University (November 1979). ROD = DhL’/v. The references for the Table 1 data used
12. S. Mochizuki and Y. Yagi, Characteristics of vortex different definitions for A and/or A,, which affects the
shedding in platearrays, Flow Visualization II, Proc. 2nd detinition of &, and Re. References [3-51 define A, = sh.
Int. Symposium on Flow Visualization, Bochum, W. References [3,5] (surfaces 1-16 and 1820) use the same .4 as
Germany, pp. 999103 (1980). in the present work. Reference [4] (surfaces 17, 21) neglects
13. W. M. Kays, Compact heat exchangers. In AGARD thearea ofthe blunt fin faces. Thedata presentedinrefs. [3-51
Lecture Series No. 57 on Heat Exchangers (Edited by were adjusted to use the Dh and Redefinitions ofequations (7)
J. J. Ginoux), AGARD-LS-57-72 (1972). and (8) respectively.
84 H. M. JOSHI and R. L. WEBB

TRANSFERT DE CHALEUR ET FROTTEMENT DANS UN EXCHANGEUR

DE CHALEUR A BANDE-AILETTE OFFSET

R&m&On presente des modeles analytiques pour estimer le coefficient de transfert thermique et le
coefficient de frottement d’une surface d’tchangeur de chaleur avcc bande-ailette offset. On considtre les
deux regimes d’tcoulement laminaire et turbulent. BasCe sur les conditions dans le sillage, une equation
est ttablie pour predire la transition entre laminaire et turbulent. Des visualisations d’ecoulement sont
conduites pour identifier la structure de l’ecoulement a la transition. La condition predite par l’equation
correspond a l’apparition de vitesses oscillantes dans les sillages de l’ailette. Des equations sont obtenues
pour le nombre de Nusselt et pour le coefficient de frottement en ecrivant les bilans d’tnergie et de quantiti
de mouvement sur une cellule unitaire de la geometric consider&e. On utilise une solution numerique pour
calculer Nu etfsur les ailettes dans le regime laminaire, et on utilise une approche semi-empiriquepour le
[Link] risultatsdecalcul sontcomparbsauxdonniesexptrimentalesprisesdanslaprCsente
etude etaussi pour des ichangeurs de chaleur [Link] modeles predisenttouteslesdon&es a mieux que

WARMEUBERGANG UND DRUCKABFALL IN EINEM WARMEAUSTAUSCHER MIT

VERSETZTEN, STREIFENFORMIGEN RIPPEN

Zusammenfassung-Die Abhandlung stellt ein analytisches Model1 zur Bestimmung des Warmeiibergangs-
Koeffizienten und des Widerstandsbeiwertes eines Warmetauschers mit versetzten, streifenfiirmigen Rippen
var. Es werden zwei Stromungsarten, laminar und turbulent, unterschieden. Aufbauend auf den Ver-
haltnissen im Nachlauf der Rippen wird eine Gleichung entwickelt, urn den Ubergang von laminarer in
turbulente Stromung zu bestimmen. Es werden Experimente zur Sichtbarmachung der Striimung durch-
geftihrt, urn die Stromungsstruktur beim fjbergang zu bestimmen. Die mit der Gleichung fur den Ubergang
ermittelte Bedingung stimmt mit dem Auftreten von Geschwindigkeits-Oszillationen in der Wirbelschleppe
iiberein. Es werden Gleichungen fur die Nusselt-Zahl und den Widerstandsbeiwert durch Erstellen von
Energie- und Impuls-Bilanzen an einer Elementarzelle der versetzten Rippengeometrie aufgestellt. Im
laminaren Bereich wird fiir die Berechnung von Nu und f eine numerische Losung, im turbulenten Bereich
eine halbempirische Naherung angewandt. Die Berechnungen werden mit Messungen an vergrogerten
Objekten, welche in der vorliegenden Arbeit durchgefilhrt wurden, und mit Messungen an Warmetauschern
der praktischen Anwendung verglichen. Der relative Fehler war stets kleiner als *20%.

TEIIJIOIIEPEHOC M TPEHHE B TEfIJIOO6MEHHHKE CO CMEIIJEHHbIMH

IIJIACTHHYATbIMH PEBPAMH

AmioTaHHn-llpennoaeHbr aHanHTHHecKHe MonenH [Link] pacHeTa K03@&iuHeHToB TennonepeHoca H

TpeHHn Ha nOBepXHOCTH TenJIoO6MeHHHKa CO CMemeHHbIMH IIJIaCTHHYaTbIMHpe6paMu. YCTaHOBneHO
Ha,IHWenByXpeEHMOBTeYeHHR:,IaMHHapHOrO HTyp6yneHTHOrO.C HCnOJIb3OBaHHeMyCJIOBHir 8CJIene
BbIBcneHOypaBHeHHennnpaC'IeTa nepeXO~aOTnaMHHapHOrOK Typ6yJIeHTHOMyTe'ieHHlO.&mOJnieHbl
3KcnepHMeHTbr no mi3yami3anmi TeYemia nnr 0npeJrenenuR CT~YKTY~L.InepexonHor0 pexoibfanoToKa.
YCnOBHe, paCCWTaHHOe n0 ypaBHeHHI0 nepeXOna,COOTBeTCTByeT yCJIOBHI0B03HHKHOBeHHR ITynbCanHfi
CKO~~CTII a cneae pe6pa. nonyHeHbI ypaeHeHHa nna pacHeTa 3HaueHafi wcna HyccenbTa H Ko3@&i-
uHeHTa TpcHHK 1136aJIaHCoa 3HeprHH H KOJIHYeCTBa L(BuXeHH,l, 3anHCaHHbIX LlJIli WIeMeHTapHOH KHetiKH
KOH@irypaIIHH CO CMemeHHbIMH IIJIaCTHH'IaTbIMH pe6paMri. ,@a paC'ieTa Nu H fnpH JIaMHHapHOM
o6reKamiu pe6ep ucnonbsyercn YriCneHHOe pemeaue, a npu Typ6yJIeHTHOM 06TeKaHHH-
nony3Mnapa9ecrorH nonxon. Pesynbrarbr pacrera cpamiuearoTcn c ~annbrMri, nonyvennbrMa B 3rok me
pa6ore Ann KOH+irypaIHiti ysenareHHor0 MacmTa6a,u c .naHHbIMH AJIK 06bIHHbIX TenJI006MeHHHKOB.
TOSHOCTbpacYeTa no Monenm next ~npenenax +20%.