Cleanabilty study of a scraped surface heat exchanger

BLEL Walid

a,(1)

, LEGENTILHOMME Patrick a, BENEZECH Thierry b and FAYOLLE Francine a

GEPEA- UMR CNRS 6144, Saint- Nazaire, France(patrick.legentilhomme@univ-nantes.fr;

francine.fayolle@oniris-nantes.fr)
b

INRA UR 638, Villeneuve dAscq, France (thierry.benezech@lille.inra.fr)

(1)
Present address: LIMAT B - EA 4250, Pontivy, France (Walid.Blel@univ-ubs.fr)

ABSTRACT
Scraped Surface Heat Exchangers (SSHE) are widely used industrially for treating food products with high
viscosity (cheese, ice cream). In this kind of heat exchangers, the presence of spinning blades allows a
regular renewal of the product surface, preventing clogging and improving heat transfer. However, if the
body of the heat exchanger is relatively easy to clean, the entry and the exit of SSHE are poorly cleanable
due to their particular geometry and the presence of seals which make cleaning difficult. A specific study was
conducted in the entry bowl of a SSHE, whose design has been optimized by the manufacturer in order to
minimize the risk of deposit eliminating the hydrodynamical dead zones. The main goal of this study is to
extend the correlation previously identified on simple systems between the wall velocity gradient and the
cleaning ability to optimize the hydrodynamic conditions leading to efficient cleanability.
For this purpose, measurements of wall shear stress were made using the electrochemical method on the one
hand, for different hydrodynamic conditions and especially in the presence of a pulsating flow. On the other
hand, cleanability measurements after microbiological fouling and cleaning were performed. Two types of
contamination were tested: spore-forming bacteria and biofilm. The clean in place protocol (0.5% NaOH,
60 C, 10 min.) is deliberately low in order to compare the different areas and allow an effective numbering
of the colonies after culture. It follows that the geometry of the bowl tested presents no dead zones. However,
the available space for flow significantly reduces the Reynolds number and turbulence intensity, which
induces three areas of increasing contamination, corresponding to conditions of mean shears and low
fluctuations. The use of a pulsating flow increases these fluctuations, and thereby reduces the residual
contamination.
Keywords: cleanability; scraped surface heat exchanger (SSHE); flow; biofilm

INTRODUCTION

The cleaning and disinfecting of food processing equipment are very common and costly in terms of time and
cleaning products. Despite progress in the field of cleaning closed surfaces, this operation is still difficult for
some pieces of equipment having specific geometry, which results in serious health risks due to the
contamination of manufactured products by the equipment itself. Scraped Surface Heat Exchangers (SSHE)
belong to this range of equipment presenting such a specific geometry. They are widely used industrially for
the processing of food products with high viscosity (cheese, ice cream, ...). However, their complex geometry
and the presence of seals at the shaft of rotation make cleaning difficult. Previous studies [1,2] have
confirmed by numerical simulations validated by PIV, that the flow in the core part of the heat exchanger is
turbulent and promotes the dropout of any deposits. However, in the entrance area, a classical geometry of
the bowl exhibits dead zones whatever the operating conditions used. This is why the Alfalaval company
chose to develop an innovative geometry of the entry and exit areas in order to minimize these dead zones.
Meanwhile, it emerged for simple geometries [3, 4] that it was possible to correlate the wall velocity gradient
and cleanability. These works have also highlighted the contribution of the fluctuating value of the wall
shear, in addition to the average value in the wrenching of spores adhering to the wall of tested devices. The
beneficial role of fluctuations in wall shear rate on cleaning was used to test other types of unsteady flows
such as the pulsating flow. Some work on straight pipes [5, 6] showed that the pulsed flows in a turbulent
regime allow an increase of mean and fluctuating shear rates and a subsequent improvement of the kinetics of
cleaning of spores. The effect of this type of flow on the pullout of biofilms adhering to the input of an SSHE
is analyzed in the present study.

The aim of our work is to study specifically the entrance area of a SSHE to extend the correlation between
wall shear forces and cleanability, identified previously on simple systems and then optimize hydraulic
conditions leading to an efficient cleaning. At this level, tests with pulsed turbulent flows with different
frequencies and amplitudes of pulsation are carried out to verify the effect of unsteady flows on complex
geometries cleanability.

MATERIALS & METHODS

1- Scraped surface heat exchanger and tested flow
Figure 1 shows a simplified diagram of the Scraped Surface Heat Exchanger (SSHE) from Alfalaval used in
the present study.

Figure 1: Schematic diagram of the studied Scraped Surface Heat Exchanger (SSHE).

The wall velocity gradients are measured using electrochemical sensors constructed from platinum wires
flush-mounted into the risky areas of the entry bowl to be flush with the flow at the level of risk areas (Figure
2). Given the influence of the wall viscous sublayer on the accuracy of electrochemical measurements in
terms of fluctuations [7], two probe diameters were used: probes 400 m in diameter which have the
advantage to allow a good precision on the fluctuating value of the shear wall and probes 1 mm in diameter,
which in turn provide better accuracy on the mean shear rate value. The pulsations of the flow are generated
by a solenoid system that can provide high pulse amplitudes and frequencies while maintaining a turbulent
flow regime [5, 6]. This system allows an harmonic variation of the pulsed flow between a minimum and a
maximum value at a given pulsation frequency. The conditions tested in this study, including in this order
[minimum flow (L h-1) - maximum flow (L h-1) - Pulse frequency (Hz)] are the following: condition a [600
L h-1 - 2200 L h-1 - 2.5 Hz] and condition b [300 L h-1 - 1500 L h-1 - 2.5 Hz] for pulsed flow and two flowrates in steady flow (2200 L h-1 and 3720 L h-1). These flow conditions have been applied for both types of
tests (measurement of wall friction and analysis of cleanability).
2- Cleaning in place procedure
The determination of residual contamination is done by cleaning up the circuit previously infected. This
contamination occurs under dynamic conditions to be closer to real conditions of contamination of food
production lines. Two types of contamination were tested, the first one to study the adhesion of bacterial
spores on the surface of the heat exchanger, the second for the analysis concerning the appearing and the
growth of biofilms on the investigated surface. In the first case, the contamination is made with milk
containing spore-forming bacteria at a concentration of 104 UFC mL-1. This milk is circulated throughout the
system during one hour. The formation of bacterial biofilms is, in turn, obtained from the circulation, during
three days, of milk contaminated with the bacteria Bacillus cereus (105 UFC mL-1) with a renewal of the
contaminating solution each day.
The cleaning process takes place under mild conditions in order to keep a residual contamination that
would demonstrate the effect of geometry and flow on the cleanability of the equipment. The cleaning in
place is done using sodium hydroxide (0.5% w / w) for 10 min at 60 C. The flow conditions, stationary and
pulsed, are applied to this level of cleaning. A prewash phase for 5 min with a flow rate of 300 L h-1 is
applied before and after this step. The detection of residual contamination has been carried out through the
casting method with nutrient agar containing triphenyltetrazolium chloride [8, 9]. Small colonies appear in
red, easy to distinguish from each other and directly accountable at the agar surface. This method allows
localizing and quantifying the residual contamination of the heat exchanger.

3- Measurement of the wall velocity gradients

The measurement of wall velocity gradients is performed by an electrochemical method, the principle of
which has been widely addressed in the literature and will not be developed here [3, 5]. This method allows
the experimental determination of the liquid-solid mass transfer coefficient which, by analogy with
momentum transfer, can be used to obtain the wall velocity gradient [10].
The electrochemical solution used is a mixture of potassium ferricyanide (3 x 10-3 mol L-1), potassium
ferrocyanide (5 x 10-2 mol L-1) and sodium hydroxide (0.5 mol L-1) used as support electrolyte. To maintain
the same flow regimes that when testing contamination and cleaning, flow-rates have been recalculated
taking into account the physical characteristics of the cleaning solution (density = 1028 kg m-3 and viscosity
= 0.985 x 10-3 Pa.s at 22 C). For a temperature of 22 C, the diffusion coefficient of ferricyanide ions in the
solution is equal to 3.65 x 10-10 m2 s-1. Obtaining the wall velocity gradient from the diffusional limiting
current distribution is done by applying the analytical solution of Sobolik et al. [11], which consists in a
correction of the quasi-stationary solution (QSS), considered as invalid for unsteady flows with high
frequency fluctuations. The mean shear rate multiplied by the viscosity of the fluid yields the mean wall
shear stress, w, which will be presented. Figure 2 presents the shape of the inlet bowl as well as the
microprobes positions.

38 37 36 35 34

Microprobes

33
32
31
30
29
28

17

18
12

11

10

22

16

15

13

14

21

27

20

19

26
25 24 23

22
21
20

19

Figure 2: position of the electrochemical microprobes

RESULTS & DISCUSSION

The electrochemical measurements were made in different zones of the entrance bowl of the heat exchanger.
The results shown in figure 3 describe the evolution of the mean local shear stress. They also allow a
comparison between the two flow conditions, steady flow of 2200 L h-1 and pulsed flow [300 L h-1 - 1500
L h-1 - 2.5 Hz]. This condition is characterized by the set-up of a recirculation zone at the center core
characterized by velocities in the opposite direction of the main flow. This last point emphasizes the effect of
the pulsations on the mean shear stress values.
Works already done on straight lengths [5] have shown the beneficial effect of pulsations on the mean and
fluctuating wall shear stress. Figure 3 shows that, despite an average Reynolds number in the presence of
pulsations significantly lower than that corresponding to the stationary condition (10,500 and 25,500
respectively for pulsed flow condition and steady state one, calculated for a diameter equal to 4.7 10-2 m), the
average stress values are very close, even greater, according to the positions of measurement, in pulsed
condition. This result confirms the observations of Blel et al. [5] on straight pipes and shows that even for
more complex geometries, the effect of pulsations is not limited to the flow core but is still dominant near the
wall. We also notice a similar evolution of the mean shear stress with the same profile for both flow
conditions tested. Thus, the pulsations applied to the flow only act on the magnitude of shear forces and do
not alter the distribution of these forces near the wall. This result can be explained by the homogeneity of
pulsation cycles (same frequency and amplitude) during the measurement period. The same remark can be
made regarding the further condition of pulsation [600 L h-1 - 2200 L h-1 - 2.5 Hz]: Figure 4 shows the rate of
increase (%) of the mean shear stress with pulsations compared that obtained in steady state. However, unlike
for straight pipes, this improvement depends on the measurement position and therefore the geometry
available to the flow. Indeed, figure 4 shows that the effect of pulsations is greater on the closed surfaces

where the flow is limited by the walls of the geometry. Further trials are planned to study the effect of cycles
of heterogeneous parameters of pulsations (amplitude and frequency) on the improvement of the mean shear
stress.
1,4

1,4
[300 L/h - 1500 L/h - 2,5 Hz]

1,2

[300 L/h - 1500 L/h - 2,5 Hz]

Steady flow (2200 L/h)

1,2

11

0,8

w (Pa)

w (Pa)

Steady flow (2200 L/h)

1,0

1,0

0,6

12

0,4
0,2

14

0,4

16

0,2

10

15

13

17

0,0

0,6

18

0,8

0,0

10
Probe number

12

15

14

16
Probe number

18

20

1,2
[300 L/h - 1500 L/h - 2,5 Hz]

2,0
[300 L/h - 1500 L/h - 2,5 Hz]
Steady flow (2200 L/h)

1,0

1,6

 w (Pa)

0,8

1,2
w (Pa)

Steady flow (2200 L/h)

20
23

0,8

22

0,6
34

24
27

25

30

0,2

31 32

0,4

26

29

0,0

19

28

0,0

18

20

36

0,4

22
24
Probe number

26

28

30

37

35
33

32
34
36
Probe number

38
38

40

Figure 3 : Evolution of the mean shear stress for different locations in the inlet bowl.
Comparison between steady state flow and pulsed one.

Regarding the evolution of local wall shear stress, the line of electrochemical probes following the direction
of the flow (probes 1-12 in figure 2) presents a significant variation between maximum and minimum values.
Probe 1, located in the tubular part of the entry bowl, exhibits the greatest wall shear stress value. The
decrease observed for the following probes can be explained by the opening of the flow section. The
curvature of the geometry at the end of this line explains the increase of the shear stress for probes 11 and 12.
Indeed, the flow which tends to follow a straight path should move towards the outside of the bend under the
effect of centrifugal forces. This results in larger friction velocities on the outside part of the bowl. The same
assumption can explain the increase of the mean shear stress on probe 18 with respect to other sensors (13,
14, 15, 16 and 17).
Observation of the stress values measured on the probes 19-27 shows an asymmetric flow due to centrifugal
forces from the curvature. Indeed, the simultaneous analysis of the location of the probes in the bowl and the
shear stress showed a low value on the row of sensors corresponds to a high shear stress on radially opposite
probes, confirming the conservation of momentum in the flow section. This entire tubular portion of the
entrance bowl presents a good cleanability level, especially using pulsation condition [300 L h-1 - 1500 L h-1 2.5 Hz]. Instead, the flat part of the entry bowl appears less cleanable.

150

11

Increasing rate (%)

120

14

90

16
9

60

10
20

30

2
4

6 8

12

13

15 18

27

31

22 24
32

23

17

37
29
30

26
0

-30

25

10

19 20
Probe number

30

33
34

36 38
40
35

Figure 4: Mean shear stress increase for the pulsating condition [600 L h-1 2200 L h-1 2,5 Hz] compared with the
steady state flow.

Figure 5 shows the colonies of Bacillus cereus spreading over the entire surface of the bowl. Low shear stress
values can explain this result (probes 29-38). The comparison between the two flow conditions presented in
figure 4 (a and b) shows a better cleanability using a pulsating flow, although very similar mean shear stress
values are observed. Indeed, figure 3a shows a less dense distribution of colonies. This result is explained by
the effect of pulsations on the fluctuating value of the shear stress that has already been highlighted in a
previous study [5,6].

(a)

(b)
Figure 5: Comparison between the residual contaminations after the cleaning in place process
(a) with a pulsed flow (b).with a steady state flow

CONCLUSION

The study of the cleanability of the entry bowl of a Scraped Surface Heat Exchanger (SSHE) emphasized the
effects of pulsed flows in the slip weakening and bacterial biofilm adhering to the wall geometry. The
electrochemical measurements have shown the beneficial effect of pulsations on the mean value of wall shear
stress for closed surfaces where the flow is confined between the walls of the geometry. For open surfaces,
the effect of pulsations is less evident. The use of cycles with heterogeneous combinations of amplitudes and
frequency pulsations could induce an increase in shear forces and subsequently improve the cleanability.
REFERENCES
[1] Yataghene M., Fayolle F., Legrand J., 2007 ; Numerical and experimental investigation of flow patterns in scraped
surface heat exchangers, ECCE6, 6th European Congress of Chemical Engineering, Copenhague, 16-20 september.
[2] Fayolle F., Yataghene M., Belhamri R., 2008 ; Hydrodynamics study during cleaning of scraped surface heat
exchangers, ICEF 10, International Congress of Engineering and Food, Vina del Mar (Chili), 21-24 april
[3] Lelivre C., Legentilhomme P., Gaucher C., Legrand J., Faille C., Bnzech T., 2002 ; CIP: effect of local wall shear
stress variation on bacterial removal from stainless steel equipment. Chem. Eng. Sci., 57: 12871297.
[4] Blel W., Bnzech T., Legentilhomme P., Legrand J., Le Gentil-Lelivre C., 2007 ; Effect of flow arrangement on the
removal of Bacillus spores from stainless steel equipment surfaces during a cleaning in place procedure. Chem. Eng.
Sci., 62: 37983808.
[5] Blel W., Le Gentil-Lelivre C., Bnzech T., Legrand J., Legentilhomme P., 2009a ; Application of turbulent
pulsating flows to the bacterial removal during a cleaning in place procedure. Part 1: Experimental analysis of wall
shear stress in a cylindrical pipe. J. Food Eng., 90: 422-432.
[6] Blel W., Legentilhomme P., Bnzech T., Legrand J., Le Gentil-Lelivre C., 2009b ; Application of turbulent
pulsating flows to the bacterial removal during a cleaning in place procedure. Part 2: Effects on cleaning efficiency. J.
Food Eng., 90: 433-440.
[7] Deslouis C., Huet F., Robin S., Tribollet B., 1993 ; Spectral analysis of wall turbulence with photolithography devised
electrochemical probe. Int. J. Heat Mass Transfer, 36 (3): 823829.
[8] Hsmark U., Bnzech T., Faille C., Rnner U., 1999 ; Bacillus spores and moulding with TTC AGAR: A useful
method for the assessment of food processing equipment cleanability. Biofouling, 14 (1): 1524.
[9] Bnzech T., Lelivre C., Membr J.M., Viet A.-F., Faille C., 2002 ; A new method for in-place cleanability of food
processing equipment. J. Food Eng., 54: 7-15.
[10] Reiss L.P., Hanratty T.J., 1963 ; An experimental study of the unsteady nature of the viscous sublayer. AIChE J., 8:
154160.
[11] Sobolik V., Wein O., Cermak J., 1987 ; Simultaneous measurement of film thickness and wall shear stress in wavy
flow of non-Newtonian liquids. Coll. Czechoslovak Chem. Com., 52: 913928.