Chemical Engineering Science, 197 I, Vol. 26, pp. 1977- 1986. Pergamon Press.

Printed in Great Britain

The mechanism of power consumption in a Votator”f-type

scraped-surface heat exchanger
A. M. TROMMELEN and W. J. BEEKS
Unilever Research Laboratories, VlaardingenlDuiven, The Netherlands

(First received 22 June 1970; in revisedform 15 January 197 1)

Abstract-In scraped-surface heat exchangers (SSHEs) the power consumed by rotating the shaft with
blades can be high, in which case the effective heat transfer rate will be reduced. A study has been
made to ascertain the mechanisms governing power consumption and to determine the influence of
design and operating conditions on its magnitude.
Power consumption was measured under various operating conditions of a SSHE, by recording the
torque and the rotational speed of the shaft. For most stirrers the power number, PO, is a unique func-
tion of the rotational Reynolds number, Re,. This is not valid for SSHEs. For the same value of Re,,
PO increases with decreasing viscosity. This is caused by two phenomena, viz:
Due to viscous dissipation the temperature of the liquid between the edge of the blade and the
wall increases and therefore the viscosity decreases.
The clearance between the edge of the scraper blade and the tube wall is dependent on the operating
conditions and the physical properties of the fluid.
A model incorporating these effects has been derived for the power consumption. The parameters
in this model have been determined from the experiments. The values of the parameters thus deter-
mined were in agreement with the results of the flow studies. An empirical equation for power con-
sumption, which is more readily evaluated, has also been derived.

INTRODUCTION showed that there is no unique relationship

WHEN viscous liquids are cooled or heated in a
scraped-surface heat exchanger of the type that
is completely filled with the liquid (further refer-
red to as SSHE), the power consumption can be
very high. When the SSHE is used as a cooler,
the power dissipated in the liquid can be close to
the amount of heat removed, thus reducing the
heat transfer efficiency. Therefore various
authors have studied the power consumption.
Skelland and Leungll] measured power
consumption during the cooling of glycerol/
water mixtures under various conditions. They
correlated their measurements by means of the
power number, the Reynolds number and the
number of rows of blades.
PO = 77 9500 ReR-1’27no’5g.
Dinglinger 121found a similar correlation.
In 1966 Trommelen and Boerema[3]
between PO and ReR; at constant ReR, PO de-
creases with increasing viscosity. This effect
was explained by a temporary increase of the
temperature of the liquid between the edge of
the scraper blade and the tube wall. In their
correlation the authors did not distinguish
between the two parts of the power consump-
tion, viz. the dissipation in the bulk of the liquid
in the annular space and the dissipation caused
by the scraping of the blades. Furthermore, it
was assumed that the heating in the liquid bet-
ween the edge of the blade and the tube wall is
adiabatic.
In more recent work Trommelen[4] derived a
correlation in which the two parts of the power
consumption were considered separately. It is
(1) then possible to compare the power consumption
in the annular space with that calculated from the
flow pattern that is described in a previous publi-
cation [51. It was further shown that the heating

tTrade name of the Chemetron Corporation, Louisville, Kentucky, U.S.A.

SAuthors’ address: Olivier van Noortlaan 120, Vlaardingen, The Netherlands.

1977
 A. M. TROMMELEN
of the liquid between the edge of the blades and
the tube wall is not adiabatic. This phenomenon
was treated in two steps, firstly, an analysis was
made of the heat transfer mechanism in the
cylinder wall and, secondly, the temperature
distribution in the liquid near the blade edge was
calculated.
In the above-mentioned work the assumption
was made that the clearance between the edge
of the scraper blade and the wall is constant. In
this paper, this assumption will be abandoned. A
balance of forces acting on the blade will be
made to calculate the clearance between blade
edge and tube wall as a function of the operating
conditions. The parameters in the equation will
be calculated from experiments.
EXPERIMENTAL
The equipment used for the power consump-
tion measurements has been described earlier
[5,6]. The SSHE, with a length of O-46 m and an
inside diameter of 76 mm, was equipped with one
of four shafts, having diameters of 46,56,62 and
68 mm. The shaft speed, which was measured
with an electric meter, was varied between 4-O
and 33.3 rev. see-‘. Two, four or six rows of
scraper blades were applied. The torque applied
to the drive shaft of the SSHE was measured
with a momentmeter (Staiger-Mohilo).
Three different glycerol/water mixtures,
having viscosities at the operating temperatures
of 0.1, 0.5 and l-5 N.sec.mmz, were used as
working fluids. The fluid was circulated through
the uncooled SSHE and the temperatures
inlet and outlet were measured. To calculate the
power input to the shalt, the torque and the
shaft speed were recorded. Part of the power was
dissipated in the bearings of the rotating shaft.
This loss was determined by measuring the
torque required to keep a bladeless shaft rota-
ting in fluids of different viscosities, thus allowing
a correction to be made to the experimental
results. The accuracy of the torque meter, with
measuring ranges of O-O-3 kg.mz.sec2, O-l
kg.m2.sec2 and O-3 kg.m2.sec2, was 2 per cent
of the full scale, i.e. O-06 kg.m2.sec2. The error
in the net torque (without bearings) is there-
and W. J. BEEK
fore O-12 kg.m2.sec2. Measurements with a net
torque of less than 0.3 kg.m2.sec2 were dis-
regarded.
It was shown previously[3] that although the
viscosity decreases over the length of the SSHE
due to the viscous dissipation a mean viscosity
can be defined (if no cooling is applied) which
enables one to calculate the power consumption;
this mean value of 71is approximately equal to
the viscosity at the arithmatic mean of T1 and
7.2.
As the mass flow rate has no influence on the
power consumption [3], all experiments were
carried out at constant mass flow rate, viz.
0.33 kg.sec-‘. The pressure does not influence
the net power consumption either (apart from
seal and bearings). Therefore atmospheric
pressure was applied.
Experiments were performed in which the
shaft speed, the number of rows of scraper
blades and the fluid viscosity were varied. De-
tailed results are given elsewhere [73. The results
of the experiments with the 56mm shaft with
six rows of scraper blades are plotted in Fig. 1.
THEORETICAL MODEL FOR POWER
CONSUMPTION
The power consumption in a SSHE will be
considered in two parts, viz. the dissipation in
the annular space and the loss due to the scra-
ping of the blade against the wall. A model will
now be derived for both parts of the power
consumption.
at
Power consumed by the scraping of the blades
If it is assumed that the edge of the blade is
rectangular (as in Fig. 2), that isothermal
flow occurs and that s 9 8, the velocity profile
between scraper blade and tube wall is linear.
The rate of shear will be rd,N/G; the power
consumed to overcome this friction can be
derived as:
PO = 9 ReRmln. (2)
Equation (2) suggests a unique relationship
1978
 The mechanism of power consumption
PO
Fig. 1. Relationship between power number PO and Reynolds
number ReR for various viscosities (d. = 56mm, n = 6).
Viscosity I) (N.s.m.?): O- 1.1; O-0.4; A-0.1.
WdtN
Fig. 2. Flow between scraper blade and tube wall.
between power number and Reynolds number.
However, Fig. 1 reveals that this specific rela-
tionship, valid for many types of stirrers, cannot
directly be applied to predict the power con-
sumption in the SSHE.
As the essential difference between a SSHE
and a stirred vessel lies in the presence or ab-
sence of scraping, the deviating behaviour of the
SSHE is probably caused by the scraping. Two
factors may cause this phenomenon, viz.
The viscous heating of the liquid between the
edge of the scraper blade and the wall is
considerable. This results in a temporary
1979
C.E.S. Vol. 26 No. 12-B
decrease in the viscosity of this liquid layer.
The clearance between the edge of the
scraper blade and the wall of the heat trans-
fer tube may not be constant, but be dependent
on the operating conditions.
The influence of both effects will be consid-
ered in the following sections.
(a) Viscous heating of the liquid between the
edge of the scraper blade and the tube wall. The
heat dissipated in the liquid is partly transported
into the scraper blade and partly into the wall
of the heat transfer tube. The heat flow into the
scraper blade is stationary, because the edge of
the blade is constantly in contact with the layer
of hot liquid near the cylinder wall. The heat flow
towards the cylinder wall is different because the
contact time of the wall with the hot liquid layer
is very short. Therefore the problem of viscous
heating at the edges of the scraper blades has
been treated in two steps:
(1) Analysis of the heat transfer mechanism
between the liquid and the cylinder.
(2) Analysis of the temperature distribution in
the liquid.
For the analysis of the heat transfer mechan-
ism between liquid and cylinder we have to
calculate the temperature distribution in the
liquid layer between blade edge and tube wall.
This is possible if the conditions at the liquid/
tube wall boundary are known. This problem can
be analysed by assuming a heat transfer resist-
ance of I /a * in the tube wall. (II* can be evaluated
by calculating the temperature profile in the tube
wall when a plane heat source is moving with
constant velocity along the surface of a semi-
infinite body.
The result is [4]:
(y* = ,.35/F (3)
For the temperature profile in the liquid layer
and in the blade differential equations can be
derived and solved[4]. Combined with the rela-
tion between viscosity and temperature:
r) = r),exp (-a’T) (4)
 A. M. TROMMELEN and W. J. BEEK

(which for small temperature ranges is in fact now be used to introduce 7*/q into Eq. (2).
q= q. exp (A/T)) the decrease in viscosity of Equations (5) and (7) can be combined to give
the liquid between blade edge and wall can be
calculated [4]: f=exp
-r2a’q*N2dt2
37 A
r)*
-= exp
rl

This is a transcendental equation that cannot be

solved analytically. For 0.1 < 5 < 2, the function
(b) Variable clearance between the edge of the 4-t 5 can be approximated by 1.25% as is
scraper blade and the tube wall. .Penney and shown in Fig. 3. If this approximation is substi-
Bell[8] suggest that the clearance between the tuted in Eq. (8) and the logarithm is taken, the
edge of the scraper blade and the tube wall is result is:
not constant, but is dependent on the operating
conditions of the SSHE. The centrifugal force ln(It;T)=-(T)“’
and the viscous force, which tend to press the
blade against the wall, are in equilibrium with the
hydrodynamic lift force caused by the slipper x
bearing effect.
The centrifugal force is proportional to m’N2dt
and the viscous force to qNd,. The hydronamic or:
lift force [9] is proportional to
(10)

We assume that the solution of Eq. (10) can

be approximated by the equation:
Under equilibrium conditions,

$_--- 1 0.4 + 1 . (11)

q*!2s2dt = pqNdt -I yN2m’dt (6) ?1 K&+1 >

3r
is valid, in which p and y are constants.
Equation (6) can be rewritten in the form:

SAA-)*
s= (7)
dyNm’+pq’

(c) Combination of effects of viscous heating and

variable clearance. The decrease in viscosity of
the liquid between the edge of the blade and the
tube wall has been expressed as a function of
various parameters (Eq. 5), in which S is still O.3(J
unknown, The value of S has been calculated as b
a function of the operating conditions and the Fig. 3. 1.2.5d/5 (curve B) as an approximation of 5+4 (curve
reduced viscosity (Eq. 7). These equations will A)forO-1 < 1 < 2.

1980
 The mechanism of power consumption

This can be written in the form: *r A/

f=&+). (12)
0.8 -

In Fig. 4, E is plotted vs. l/(7*/7) - 1 in the range

0.4 < r)“‘/q < 1 both with Eqs. (10) and (12).
From this plot it follows: K = 0%. Eq. (9) E
then gives:

/ I I
0.1 l I I
(13) 0.1 0.2 0.4 0.8 I 2

I
Equation (2) for the power consumption can --I
“7,
now be used if Eq. (7) is substituted for 6 and
Eq. (13) for the viscosity in the Reynolds number: Fig. 4. Approximation of Eq. (10) (curve A) by Eq. (12)
(curve B).

When there are no blades, the velocity profile is

linear. After some calculations it is found:
or:
2m3dt
PO = dt _ d, ReR-’ . (17)
y(Nm’h) +P
PO = ntiReRvl
1 + qN1.75 [y (Nm’lv) -t pl”.25y
(14) From Eqs. ( 16) and ( 17) it appears that the power
number is inversely proportional to the Reynolds
With:
number:
7.2a’dl.75
y = hO~5(sp,C,h,)0’25’ (1% PO = R ReR-l. (18)

Power consumed in the annular space Around the blades the flow pattern is compli-
For the calculation of the power consumed in cated, but this region is small compared with that
the annular space two flow regimes will be con- of the undisturbed part of the annulus. and there-
sidered, viz. laminar Couette flow and the fore Eq. (18) will still hold. The value of R can
Taylor instability regime. For the Couette flow be calculated from the velocity profile, published
regime we first consider the case where there is earlier151 for the shaft with a diameter of 56 mm.
no clearance between scraper blade and shaft. If This will be discussed below.
the annular space, i(d,-d,), is small relative to For the Taylor instability regime it has been
the diameter, the flow is similar to that between shown [lo] that for Reynolds numbers that are
two flat plates; for the power consumption in not appreciably higher than the upper limit of the
this case can be found: Couette flow regime the power consumption
may be represented by:
- t8tid ReR-1
(16)
‘O-d,-d, . PO = constant ReR-0.5. (19)
1981
 A. M. TROMMELEN and W. J. BEEK

At the critical Reynolds number, discussed of the parameters from the experimental data
previously [5], Eqs. (18) and (19) should yield the was performed as follows:
same value. This results in: It was first estimated what values could be
expected for the parameters. The ranges of the
PO = R Re,+ for ReR < Recr parameters were then divided in a number of
PO = R Re,rm05RR-o.5for Re, > Recr I - (20) equal-sized steps. At all possible combinations of
y, p, y and R that were thus obtained, the sum of
EVALUATION OF THE PARAMETERS IN THE the squares of the difference between measured
MODEL FROM THE EXPERIMENTAL DATA and calculated power number, divided by the
From Eqs. (14) and (20) it follows that the experimental power number was determined. If
power consumption in a SSHE can be described one or more of the values of the parameters that
by: corresponded to the lowest sum was on the
boundary of a range, the procedure was repeated
PO = rm2ReR- with new estimates of the parameters. If the
values of the parameters with the lowest sum
+f(Red (21) were within the ranges, the step size was re-
duced until sufficient accuracy for the values of
with:
y, & y and R was obtained. The calculations
7.-&$d 1.75
were performed with a digital computer.
(15) The results of this first step in the calculation
procedure are given in Table 1.
and:
From the derivation of the model for
f (ReR) = R ReR-’ for ReR < Recr power consumption, it follows that y and y are
f (ReR) = R Rec,-0’5ReR-o~5
for RR > Re,, I ’ not dependent on the shaft diameter and the
number of scraper blades, y is a constant that is
(22)
determined by physical properties of tube wall
There are four unknown parameters in the equa- and working fluid, the geometry of the blade
tions, viz. y, /3, y and R, that can be determined edge and the tube diameter. y is determined by
from the experimental data. As Eq. (21) cannot the ratio between the hydrodynamic lift force
be linearized it is not possibie to apply regression and the centrifugal force.
analysis to find the parameters. The calculation However, the values of y and y in Table 1

Table 1. Evaluation of y. 8, y and R in EQ. (2 1)

Shaft No. rows

dia. of scraper Y B yx1os R .kP
(mm) blades (kg-‘.m.sec*‘7s) (&

46 2 588 2920 0.40 0 130 11

46 4 540 3100 l-00 0 167 13
46 6 512 3640 2.00 0 168 13
56’ 2 788 3740 0.70 0 127 11
56 4 580 4775 1.12 30 61 8
56 6 248 4184 3.28 814 42 6
62 2 600 6200 0.50 0 47 7
62 4 376 5700 1.20 0 98 10
62 6 332 5300 1.40 0 48 7
68 2 2602 8524 3.78 1195 3 2
68 4 294 5079 1.98 1546 1 1
68 6 248 4932 1.38 1146 2 1

1982
 The mechanism of power consumption

vary considerably. This is due to the fact that Table 3. Evaluation of /3 and R in Eq. (21) with Y fixed
different combinations of the four parameters at 1-4X 10m3kg-l.m.sec*“Jand y at 380
may result in almost the same values for Sz.
Shaft No. rows
This is also the reason why values of R = 0 were dia. of scraper fi R 6
obtained in a number of series of experiments. (-1 blades (&
Hence, as a second step in the calculation, the
46 2 1937 343 1% 14
average value of y has been determined from the 46 4 2681 13
251 182
values in Table 1 (the lowest and highest values 46 6 3240 0 183 14
not being considered). This resulted in y = 1.4 X 56 2 1987 371 184 14
10e3. The calculation was then repeated with this 56 4 3835 365 67 8
56 6 5200 0 53 7
value of y (Table 2). 62 2 3684 463 72 8
62 4 5684 58 103 10
62 6 5300 0 52 7
Table 2. Evaluation of y, /3 and R in Eq. (21) withy fixed at
68 2 9429 1035 35 6
1.4 X lo-$ kg-1.m.sec2’75
68 4 6414 1135 2 1
68 6 5243 990 3 2
Shaft No. rows
dia. of scraper y /3 R Sz
(mm) blades (& Table 4. Evaluation of p in Eq. (21) with y iixed at
1.4 X lo+kp-‘.m.secp’75, y at 380 and average values
46 2 210 1888 371 193 14 for R
46 4 438 2594 233 180 13
46 6 400 3240 0 181 13
56 2 564 2622 269 164 13 Shaft No. rows
56 4 494 4317 233 62 8 dia. of scraper fi R Sz
56 6 400 5200 0 51 7 (mm) blades (&
62 2 468 4474 385 68 8
62 4 386 5924 36 102 10 46 2 2358 300 1% 14
62 6 300 5300 0 50 7 46 4 2439 300 183 14
68 2 1624 11,477 762 10 3 46 6 2276 300 228 15
68 4 269 6133 1229 1 1 56 2 1982 370 184 14
68 6 244 4930 1158 2 1 56 4 3800 370 67 8
56 6 3700 370 73 9
62 2 3735 460 71 8
62 4 3374 460 173 13
From the values of y, reported in Table 2, the 62 6 3425 460 107 10
average was 380 (the highest value being 68 2 9198 1050 35 6
omitted). Again calculations were carried out to 68 4 6939 1050 2 1
68 6 5070 1050 3 2
evaluate the two yet undetermined parameters,
P and R. y was kept constant at 1.4 x 10b3, and y
at 3 80. The results are given in Table 3. data to the model, increases only from 7.5 to
From the series of measurements that gave the 9.5 per cent.
best fit to the equation (shaft diameter 68 mm) it
appears that R does not depend on the number of DISCUSSION OF THE PARAMETER VALUES
rows of scraper blades. Therefore for each shaft Equation (15) gives the relation between y and
diameter the average value of R was taken, various physical properties and geometrical
omitting the extremely low values. The results of quantities. The only unknown factor is S, the
the calculations with only one parameter, p, thickness of the edges of the scraper blades. If
undetermined.are given in Table 4. one calculates s from Eq. (21) withy = l-4 x 1O-3
Although there is only one degree of freedom the result is: s = O-028 mm. This is a reasonable
for the calculation, reported in Table 4, against value, although it can only be taken as a rough
four for those in Table 1, the average value of S, estimate, because s is very sensitive to small
which is a measure of the fit of the experimental changes in y. For y = 1*OX 10m3Eq. ( 15) yields
1983
 A. M. TROMMELEN
s = 100 X 10e6m and for y = 2.0X 10-4, s = 6.8 x
10e6 m. Hence a twofold increase in y results in a
value of s that is 16 times smaller.
Above (Eqs. 16 and 17)R was derived and
found to be dependent on the ratio of tube
diameter and annular space, d,/(d,- d,). For the
two extreme conditions, viz. no blades at all and
completely closed blades, R is 2+dt/(d, - d.)
and 8x3d&(d, - d,). In Table 5 the values of
Table 5. The parameter R as a function
of the four shaft diameters
dt-4
Shaft R R.7
f
No blades 2+d, 62
(theoretical) d,-d,
Closed blades sad,
248
(theoretical) d,-d,
d,=46mm 300 118
d,=56mm 370 97
d,= 62 mm 460 85
d,=68mm 1050 110
R (dt - d,)/d, are given for the four shaft dia-
meters.
There is no systematic dependency of R (dt -
d,)/d, on the shaft diameter; the average is 102.
We can compare these results with the velocity
profile for the shaft with a diameter of 56 mm,
determined earlier[5]. From the velocity profile
the shear rate at the tube wall can be calculated
and from this the power consumption in the
annular space. The result for this case is:
R (dt - d,)/d, = 101. The agreement between the
measurements of the power consumption and the
velocity profile is almost too perfect.
The values of y and p are related to the equili-
brium between centrifugal force, viscous force
and hydrodynamic lift force. For a shaft speed of
10 rev. set-l the centrifugal force per meter
blade length is 35 N(m’ = 0.117 kg.m-l).
For the 56 mm shaft and a liquid viscosity of
O-1 N.sec.m+ this means that the viscous force
per meter blade length is 35PqNdt/yN2m’dt
15.5 N. The hydrodynamic lift force is there-
fore 50 N.
From Eq. (I 3) the decrease in viscosity due to
and W. J. BEEK
viscous heating can be calculated for this case.
The result is q* = O-094 N.sec.m.-2 and with
71= qOe-a’r it follows that the increase of the
temperature of the liquid at the edge of the blade
is: AT = 0.8”C. From Eq. (7) the clearance
between the edge of the scraper blade and the
tube wall can be calculated roughly as 0.5 X lOwe
m. This is only an approximation as the value of
s could not be determined accurately. However,
we may conclude that the clearance between the
edge of the scraper blade and the tube wall is of
the order of 1 X 10m6m.
It is important to realize that the foregoing
analysis was derived for an adiabatic votator.
But the apparatus is, of course, never used that
way. If heat transfer takes place, the temperature
in the bulk of the liquid is different from that near
the heat transfer tube. This phenomenon can be
accounted for by substituting the viscosity of the
liquid at the wall temperature in the part of the
power consumption due to scraping.
EMPIRICALCORRELATION FOR POWER
CONSUMPTION
From Eqs. (2 l), (22) and (15) and the values of
the parameters given in Table 4, the power
consumption can be predicted if the operating
conditions are known. To derive an equation of a
simpler form, it was assumed that the power
consumption could be written in the form:
P = u, ( Ndt)U1qu2nu3L
(dt-- d,)“d ’ (23)
The logarithm of Eq. (23) was taken and multiple
linear regression analysis was applied to the
experimental data in order to find uO. . . . . . u4.
The result was:
P = 251 (N:,:‘~~~~$68L. (in m.k.s. units).
t I
(24)
= The multiple correlation coefficient was 0.992.
An attempt was also made to see whether power
consumption measurements for another SSHE
(L = O-26 m), (dt = 0.05 m) reported previously
1984
 The mechanism of power consumption

[3] could be correlated in the same way. The was derived in the form:
values of u3 and u4 were taken from (60) and uo,
u1 and u2 were determined, resulting in: p = 251(Ndt)1’79~0’66no’gsL~. .s .
(form k u&s)

P = 276 (N~$~~‘~~~P’6’L .(in m.k.s. units). (25)
t 8’
Equation (25) is only slightly different from Eq.
(24).
When the data of Skelland and Leung on
power consumption, are compared with Eq. (24)
good agreement between these experiments and
the equation is found.
CONCLUSIONS
For most stirrers the power number is a unique
function of the rotational Reynolds number. This
is not valid for SSHEs. For the same value of
ReR, PO increases with decreasing bulk viscosity.
This is caused by two phenomena:
Due to viscous dissipation the temperature of
the liquid between the edge of the blade and
the wall increases and therefore the viscosity
decreases.
The clearance between the edge of the scraper
blade and the tube wall is dependent on the
operating conditions.
A model incorporating these effects has been
derived for the power consumption. The
equation reads:
(dt - dJ+31
Acknowledgements-The authors thank Mr. N. A. M.
Lansbergen for carrying out the measurements and Mr. P. R.
Croes for writing the computer programme.
NOTATION
a’
constant in viscosity equation, T-l
cw specific heat of wall material, L2r2 T-l
d, shaft diameter, L
dt diameter of heat transfer tube, L
K constant in Eq. (11)
L length of scraped surface, L
m’ mass of scraper blade per unit length,
ML-’
N shaft speed, r-l
number of rows of scraper blades
1: power consumed by rotating shaft plus
blades ML2r3
PO P/pN3dti = power number
R parameter in power consumption model
R% d?Np/q = Reynolds number for rotation
Recr critical rotational Reynolds number
width of scraper blade edge, L
T” temperature, T
T1, T, inlet and outlet temperature of fluid, T
Y parameter in power consumption model,
kg-’ m sec2.75.

Greek symbols
PO = nrr2ReR- 1 Jim+ p10.25y heat transfer coefficient between heat
source and semi-infinite body, Mr3T-’
+fWJ parameters in power consumption model
clearance between blade edge and tube
where
wall, L
and f(ReR) = R ReR-lforReR -C Re,,
constant, used for solving Eq. (8)
fluid viscosity, ML-‘?-’
f (ReR) = R Recr-o’5ReR-0’5 for ReR > viscosity of liquid between blade edge
Re,,. and wall, ML-‘?-’
viscosity in Eq. (4), ML-‘t-l
The parameters in the equation, y, p, y and R thermal conductivity of fluid, MLr3T-’
were determined experimentally, and their thermal conductivity of wall material,
values were in agreement with the results of the MLt-3T-’
flow studies [5]. constant in Eq. ( 10)
An empirical equation, more readily evaluated, density of wall material, MLe3
1985
 A. M. TROMMELEN and W. J. BEEK

REFERENCES
111 SKELLAND A. H. P. and LEUNG L. S., Br. Gem. Engng 1962 7 264.
[2] DINGLINGER G., Kiilterechnik 1964 16 170.
[3] TROMMELEN A. M. and BOEREMA S., Trans. Inst. Chem. Engrs. (London), 1966 44 T329.
[4] TROMMELEN A. M., Paper presented at the 3rd Cong. CHISA, 15-20 September 1969, Mari&nske L&me, Chechos-
lovakia.
[5] TROMMELEN A. M. and BEEK W. J., To be published.
161 TROMMELEN A. M.. Trans. Inst. Chem. Ennrs. (London) 1967 45 T176.
[7] TROMMELEN A. M., PhysicalAspects ofdckped-surf& Heat Exchangers, Thesis Delft, 1970.
181PENNEY W. R. and BELL K. J., Ind. Engng Chem. 1967 59 (apr.) 40.
[9] NORTON, A. E., Lubrication. McGraw-Hill, New York, 1942.
[lo] VAN LOOKEREN and CAMPAGNE N., Longitudinale dispersie in de stroming door een ringspleet met draaiende
binnencilinder, Thesis, Groningen, 1966.

Resume-Dans des echangeurs de temperature a surface balayee (ETSB), I’tnergie consommee en

faisant toumer I’axe muni de lames peut &tre importante et dans ce cas le transfert de chaleur est
reduit. Une etude a 6tt faite pour determiner les mecanismes gouvemant la consommation d’energie
et pour determiner l’influence sur sa grandeur de la construction et des conditions d’operation. La
consommation d’energie fut mesuree sous differentes conditions d’operation d’un ETSB en notant le
moment du couple et la vitesse de rotation de l’axe. Pour la plupart des agitateurs, Ie nombre PO,
representant la consommation d’bnergie, est une fonction unique du nombre de rotation de Reynolds
Re,. Ceci ne fut pas valable dans le cas des tchangeurs de temperature a surface balayee.
Pour la m&me valeur de Re,, PO augmente quand la viscosite diminue. Ceci est dfi a deux pheno-
m&es:
a cause de la dissipation visqueuse, la temperature du liquide entre le bord de la lame et la paroi
augmente et par consequent la viscosite diminue.
I’intervalle entre le bord de la lame et la paroi du tube depend des conditions d’op&ation et des
proprietes physiques du fluide.
On a determine pour la consommation d’energie, des equations tenant compte de ces effets. A partir
des experiences on a determine les parametres de ces equations. La valeur de ces paramttres s’accor-
daient avec les rtsultats des etudes de l’ecoulement. On a aussi ttabli une equation empiricme
. _ de la
consommation d’tnergie, qui est facile a tvaluer.

Zusammenfassung- In Wiirmeaustauschem mit Oberthichenschabung (scraped-surface heat exchan-

gers, SSHE) kann die bei der Rotation der Schaufelwelle verbrauchte Leistung recht hoch sein,
wodurch dann die wirksame Warmeiibertragung vermindert wird. Es wurde eine Untersuchung
durchgefiihrt urn die den Leistungsverbrauch bestimmenden Vorgange festzustellen und den Einfluss
der Konstruktion und der Betriebsbedingungen auf das Ausmass desselben zu bestimmen.
Der Leistungsverbrauch wurde bei verschiedenen Betriebs-bedingungen eines SSHE, durch
Aufzeichnung des Drehmoments und der Umlaufgeschwindigkeit der Welle, gemessen. Fur die
meisten Riihrer ist die Leistungszahl, PO, eine eindeutige Funktion der Rotations-Reynoldsschen
Zahl, Re,. Das gilt nicht fur SSHEs. Fur den gleichen Wert von ReR nimmt PO mit abnehmender
ViskositPt zu. Das wird durch zweierlei Erscheinungen hervorgerufen, nlmlich:
Infolge viskoser Verteilung nimmt die Temperatur der Fliissigkeit zwischen der Schaufelkante und
der Wand zu und die Viskositlit nimmt daher ab.
Der freie Raum zwischen der Kante der Schabschaufel und der Rohrwand hiingt ab von den Betrie
bedingungen und den physikalischen Eigenschaften der Fliissigkeit.
Es wurde ein Model1 flir den Leistungsverbrauch abgeleitet, dass diese Wirkungen beriicksichtigt.
Die Parameter dieses Modells wurden aus den Versuchen bestimmt. Die so bestimmten Werte der
Parameter stimmten mit den Ergebnissen der Strijmungsuntersuchungen iiberein. Eine empirische
Gleichung flir den Leistungsverbrauch, die eine schnellere Beurteilung gestattet, wurde ebenfalls
erhalten.

1986