4 Results
4.1 Power Drawn
The determination of the power consumed by rotating impellers in mixing tanks or reactors is important for the calculation of the energy requirement of the system. Furthermore, the knowledge of the power number for every impeller is helpful in the determination of the equivalent specific energy dissipation rate, ε
T; this parameter allows comparing different impeller. In fact, it will be necessary compare mixing time against ε
T, despite the impeller speed (see also section 4.2.2).
In this section will be shown first the results concerning the power in ungassed conditions, then the same impellers working in gassed conditions, the air was introduced directly underneath the centre of the impellers.
4.1.1 Ungassed
The device used to measure the torque transmitted from the motor to
the shaft, could be used only in vertical position. Because the shaft
should be angled at 45°, some different solutions were studied to
determine the real power consumption in the vessel. For this purpose
the three conditions shown in Figure 4-1 were analyzed. All of them
C=H/4
C=20 mm
C=20 mm 45
have the shaft in the vertical position, for the first one has been chosen a standard clearance (one fourth of the high of the liquid
over the impeller), then the second has a clearance equal to the real clearance when the impeller is angled; while for the third case an angled plane was added underneath the impeller, to simulate the deviation of the pumped flow.
The results concerning the three different power measures for 3PBT30 are shown in Figure 4-2. In this graph, it can be observed how the Po does not change substantially with modifying measuring configuration (0.38<Po<0.61). Specially with increasing Re, the Po falls within (0.40;0.45) range for Re>25000. Hence the average value of 0.42 can be assumed to be the Po of this impeller for ungassed condition and within the turbulent region. According to Rewatkar et al.
15the power number increases with a decrease of the clearance from the vessel bottom, in fact for lower impeller speeds the Po when the clearance is 20mm is 30-50% greater than C=H/4 condition. The solution with the
Figure 4-1: Configurations to determine Po
plane angled of 45
oseems to have intermediate values.
Like the previous impeller, same power measurements were done with the STAS DOWN impeller; also in this case the Po did not vary appreciably (see Figure 4-3). The value of Po that will be taken as constant in the turbulent region for this impeller is equal to 1.55. This value is really similar with the Po reported by Armenante et al.
16for pitched-blade turbine with 4 and 6 blades. Also in this case the power absorbed when the clearance is 20mm is slightly higher, but does not change significantly.
rpm
400 500 600 700 800 900 1000
Po
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Re
15x103 20x103 25x103 30x103 40x103
C=H/4 45o plane C = 20 mm
Figure 4-2: Power number for 3PBT30 in ungassed condition.
For the last impeller these different measures were not done, but, as shown in Figure 4-4, comparing the STAS UP impeller with the STAS DOWN, it can be noticed the power number is similar for upward and downward pumping. The average value of the Po in the turbulent region (with c=20mm) is 1.61.
Re
15x103 20x103 25x103 30x103
10x103
Po
0,8 0,9 1,5 2 3
1
rpm
400 500 600 700 800 900 1000
C = H/4 45o plane C = 20 mm
Figure 4-3: Power number for STAS DOWN in ungassed condition.
4.1.2 Gassed
For gassed system, according to the theory about Rushton impellers, the Po decreases with the increasing Re; this is due to the fact that behind the blades of the impeller large cavities appear, hence the impeller is acting in a two-phase fluid having a lower density.
Something very similar happens also with these kinds of impellers, in fact observing the impeller acting with the introduction of gas, some cavities were developed within the impeller. This behaviour was more evident for STAS impellers as the graphs below show. Furthermore, since all these impellers produced an axial flow, it happened that for low gas flow rate and higher rotational impeller speed the pumping capacity of the propeller was able to completely disperse the gas from the impeller, making the Po rise to the ungassed values.
Re
15x103 20x103 25x103 30x103
10x103
Po
1.5 2 2.5
STAS UP STAS DOWN
Figure 4-4: Comparing Po for STAS DOWN and STAS UP impellers.
The Figure 4-5 shows the Po and Po
gfor the 3PBT30 against Re with and without the gas presence. As mentioned above, the Po
ghas a lower value than the Po for all the impeller speeds, however, for
min / 1 . 0 l
Q
v= and speeds over 850rpm, the pumped liquid pushes away the gas from the impeller. For these speeds a higher Po
gcould be noticed in the graph. A larger gas flow rate does not change a lot the gassed power number; in fact the trend of the blue triangle is the same of the lower gas flow rate, furthermore a bigger amount of gas in the impeller does not make the impeller disengaging the gas from the blades, even for very high impeller speeds. For STAS impellers the
rpm
400 500 600 700 800 900 1000
Po or Po g
0,15 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9
0,1 1
Re
15x103 20x103 30x103 40x103
Ungassed Qv = 0.1 l/min Qv = 1.0 l/min
Figure 4-5: Power consumption for 3PBT30 in gassed situation.
behaviour is a bit different, in fact for lower gas flow rate the Po
gis minor of the Po, ungassed condition, but almost constant for all rotational impeller speed. For larger amounts of gas the direction of pumping changes the power consumption completely.
Figure 4-6 shows the Po and Po
gfor the STAS DOWN impeller versus the Re or the impeller speed. As already mentioned above, for Q
gequal to 0.1 l/min the Po and the Po
gfollow the same trend also if the Po
gassumes values minor than the Po. In fact, with the introduction of the gas, the impeller is acting in a two-phase fluid and no longer in a simple liquid, so, using the density of the liquid in the formula to calculate the Po
g, a smaller value is obtained. That means that the system absorbs a decreasing power. For Q
gequal to 1.0 l/min the Po
grpm
400 500 600 700 800 900 1000
Po or Pog
0,4 0,5 0,6 0,7 0,8 0,9 1,5 2,0 3,0 4,0
1
Re
15x103 20x103 25x103 30x103
10x103
Ungassed Qv = 0.1 l/min
Qv = 1.0 l/min (direct loading) Qv = 1.0 l/min (indirect loading) Return from 1300rpm
Figure 4-6: Power consumption for STAS DOWN in gassed situation.
decreases steadily for increasing Re. Furthermore it can be observed for 14500<Re<19000 an unstable region appears, where big cavities are developed behind the blades of the impeller, so the impeller is rotating in almost only gas. But the pumping capacity of the impeller is able to disperse the gas away from the impeller; suddenly the impeller is working again in only liquid and the absorbed power increases. To understand better this phenomenon, the impeller speed was increased till 1300rpm, when the gas was completely dispersed from the blades and the impeller was acting in only liquid, after these conditions were achieved the speed was gradually decrease and the Po
gfollowed the green line; when the pumping capacity was not enough to keep out the gas from the impeller, the power dropped to the old values.
In
Figure 4-7 are shown two images where the gas is engaged or not in
the STAS DOWN impeller.
Figure 4-7: Examples of engaging and disengaging of the gas
In Figure 4-8 is illustrated the power consumption concerning the STAS upward pumping impeller; in this case all the observation done for the STAS DOWN are still valid, the main difference is the fact that the unstable region begins for Re greater than 18500, and remains also for high rotational impeller speeds. Even increasing the impeller speed to 1500rpm was not enough to enable it to disengage the gas.
rpm
400 500 600 700 800 900 1000
Po or Po g
0,4 0,5 0,6 0,7 0,8 0,9 1,5 2
1
Re
15x103 20x103 25x103 30x103
10x103
Ungassed Qv = 0.1 l/min Qv = 1.0 l/min Qv = 1.0 l/min
Figure 4-8: Power consumption for STAS UP in gassed situation.
4.2 Mixing Time
As mentioned in section Errore. L'origine riferimento non è stata
trovata. one of the most widelyused techniques to clean the molten aluminium is the introduction of a inert/chlorine mixture using a static lance which provides the cleaning reagents and the needed mixing for the furnace.
The mixing generated by the lance is usually inadequate for the geometry and dimension of melting vessel; hence, the possibility to add an impeller has been studied. To compare the lance system with the impeller system, mixing time was investigated as a function of the specific energy dissipation rate, ε
T. Furthermore, combined solutions with different impellers and different gas injection were analysed.
To obtain the mixing time, decolourisation technique was used. The mixing time is defined as the time lapse from onset of the chemical reaction, between iodine and sodium thiosulphate in presence of starch, till a colourless solution is obtained. For every configuration, several measures were taken aiming at obtaining an average value
Figure 4-9: Decolourisation sampleof impeller system
of mixing time: this kind of geometry generates time depending flows, and sometimes quite large diverging values were obtained, or dead zones appeared making the final time hard to determinate. The number of trials was depending on the deviation of the results, from 3 to 5 measures.
4.2.1 Lance System
In this condition two gas flow rates were analysed, both were a simple scale down of industrial value, e.g. 0.1 and 1.0 l/min ( ε
Tequal at 0.1 and 1.0 W/m
3respectively). For the first gas flow rate, 0.1l/min, was impossible to determine the mixing time, because very long and completely different values were collected, the average would not be a good approximation of the real value or the error range should be much too high. On the other hand, the second gas flow rate gave a mixing time, θ
m, of 295sec.
Figure 4-10: Lance scheme of liquid flow pattern
4.2.2 Impeller System
To substitute the lance system, three different impellers were
analysed: a normal three-pitch blades turbine, downward pumping,
3PBT30; and two impellers designed by S.T.A.S. - UNIGEC 2002, which had the same size and shape, but opposite pumping, STAS-UP and STAS-DOWN. To compare the three impellers, mixing times were measured for equivalent energy dissipation rate, these are collected in Table 4-1; further details concerning their evaluation are explained in section Errore. L'origine riferimento non è stata trovata..
As shown in Figure 4-11, the direction of pumping is very important, in
Table 4-1: Equivalentε
T for different systemsε
T(W/m
3)
Lance
(l/min)
3PBT30(rpm)
STAS-DOWN(rpm)
STAS-UP(rpm)
1 1 348 141 140
2 438 303 301
4 552 382 380
7 666 481 478
10 750 580 577
25 653 649
εT (W/m3)
0,91 2 3 4 5 6 7 8 910 20
Mixing time (sec.)
20 40 60 80 100 120 140
3PBT45 STAS-DOWN STAS-UP
Figure 4-11: Mixing time for impeller systems
fact, while the STAS-DOWN and the 3PBT30 impellers show a similar behaviour, the STAS-UP indicates a worse mixing in every condition.
This is due to the geometry of the vessel, in fact, in these conditions, the part near to the impeller was well mixed, but the rest of the vessel was quite steady, and the decolourisation phenomenon was like simple diffusion, instead of being due to the flow. The downward pumping impellers work in a very similar manner, even if the 3PBT30 seems to be better for increasing energy dissipation rate
4.2.3 Combined System 4.2.3.1 Injection Point
The first issue relating to combined system was how the mixing time is influenced by a change in the injection point. For this topic, only the 3PBT30 (downward pumping) was used in several configurations.
Three configurations were studied maintaining the position of the impeller, while changing the position of the injection of the gas. In fact, the impeller was positioned with an angle of 45° from the horizontal plane, with a gap of 200mm from the left side of the vessel and of 20mm from the bottom, while the injection points were below the hub, on the periphery of the impeller and above this, as shown in Figure 4-12. All these settings were applied in two different gas flow rates, at 0.1 l/min and 1.0 l/min, and several impeller speeds.
Figure 4-12: Different injection points
A B C
During this kind of experimentation it could happen that some dead zones appeared, thus an intermediate time was usually taken as mixing time and then a final time was taken too. Furthermore, for every particular condition (configuration, gas flow rate and impeller speed), four different time measures had been taken, so the mixing time was the average of all these values.
Figure 4-13 shows the different configurations with a flow gas rate of 0.1 l/min. In this condition the solution B is the worst, in fact it has the longer mixing time compared with the other solutions. Probably the gas injection point situated directly in the periphery of the impeller decreases the liquid pumping, because the gas is completely engaged into the impeller. By contrast, two other solutions, above and below the impeller, are very similar, with a mixing time smaller of
(εT)g(W/m3)
0,91 2 3 4 5 6 7 8 9 10
Mixing Time (sec)
0 20 40 60 80 100
Configuration A Configuration B Configuration C
Figure 4-13: Mixing time 3PBT30+ Air (0.1 l/min) different injection points
about 15-20 sec. Furthermore, except for the first configuration, in almost all these experiments several dead zones appeared as zones where there was no flux; in these cases, the mixing time was taken when the whole solution was colourless, neglecting those little areas, valuated to be not more 3% of the whole vessel.
The Figure 4-14 shows the data collected using a gas flow rate of 1 l/min, the first configuration (A) is the best, while the other two show higher mixing time s of at least 20 sec. In these conditions dead zone never appeared. Furthermore, in all the three configurations the mixing time stabilizes for increasing impeller speeds.
(εT)g(W/m3)
0,9 1 2 3 4 5 6 7 8 9 1 0
Mixing Time (sec)
0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0
Configuration A Configuration B Configuration C
Figure 4-14: Mixing time 3PBT30+ Air (1.0 l/min) different injection points
Concerning Figure 4-15, this graph shows all the configurations together, with the aim to compare the same solution in different gas flow rate. Only the third configuration, C, shows a significant difference between the two gas flow rates, in fact when the gas flow rate is 0.1 l/min the mixing time is at least 15 sec less. Furthermore, the graph shows that the gas flow rate modifies the flow in the three configurations in a completely different way.
According to the mixing time the first configuration, A, is the best, while the other two show similar behaviours.
(εT)g(W/m3)
0,9 1 2 3 4 5 6 7 8 9 10
Mixing Time (sec)
0 20 40 60 80 100 120 140
Conf A Qgv = 0.1 l/min Conf B Qgv = 0.1 l/min Conf C Qgv = 0.1 l/min Conf A Qgv = 1 l/min Conf B Qgv = 1 l/min Conf C Qgv = 1 l/min
Figure 4-15: Mixing time for different injection point in different gas flow rate
4.2.3.2 Impeller Types and Gas Injection (0.1 l/min)
The previous section has shown that configuration A, concerning the mixing time, is the best. At this point the comparison between the different impellers was done using the A injection point, also because it is the easiest realizable, in fact, using a hollow shaft, the injection point is below the hub of the impeller.
Comparing Figure 4-13 with Figure 4-16, a low introduction of air (Q
v= 0.1 l/min) does not appreciably change the efficiency of the single impellers; in fact, the 3PBT30 shows mixing time comparable to ungassed system, while the STAS DOWN exhibits longer mixing time than the 3PBT30 impeller, however in ungassed conditions the two
εt,(W/m3)
0,91 2 3 4 5 6 7 8 910 20 30
Mixing time, (sec.)
20 40 60 80 100 120 140
3PBT45 STAS DOWN STAS UP
Figure 4-16: Mixing time for different impellers with gas flow rate Qv=0.1 l/min
propeller are similar. Concerning the STAS UP, this impeller with an introduction of a low amount of gas displays the same value of mixing time in ungassed and gassed condition. This has taken place because the pumping liquid and the gas had the same direction, so the natural movement of gas was helped from the liquid and passed directly through the impeller without changing its capabilities.
4.2.3.3 Impeller Types and Gas Injection (1.0 l/min)
Figure 4-17 shows the mixing time for the three different impellers in upper gassed conditions (Q
v= 1.0 l/min); the STAS UP has the same behaviour of the previous experiment with a slightly decreasing
(εT
)
g, (W/m
3)
0,91 2 3 4 5 6 7 8 910 20 30
Mixing time(sec.)
20 40 60 80 100 120 140
3PBT45 STAS DOWN STAS UP
Figure 4-17: Mixing time for different impellers with gas flow rate Qv=1.0 l/min
performance for higher impeller speed
13. On the other hand, the STAS DOWN shows a different trend, in fact, for lower impeller speed the mixing time is almost the same of the previous case, while for higher density energy dissipation rate the mixing time assumes higher values
13, while the 3PBT30 is still the best of three, in fact it has the lowest values in terms of mixing time.
4.2.4 Asymmetrical configurations
After the 3D-PIV analysis, reported in section 4.3.2, an asymmetrical study of the system was necessary. In fact, from that experiment was clear how the flow generated from the STAS impeller was not perfectly axial, but the effect of the bottom of the vessel diverted the flow. So some new configurations were analysed:
STAS_DOWN SHAFT ANGLE WITH THE BOTTOM 45°
Vessel dim 650x252x130 mm
addition point: on liquid surface over the impeller Impeller speed 480 rpm
A= 200mm B= 70mm
Mixing Time AVG STD DEV
52 58 57 55.7 3.21
STAS_DOWN SHAFT ANGLE WITH THE BOTTOM 45°
Vessel dim 650x252x130 mm
addition point: on liquid surface over the impeller Impeller speed 480 rpm
A= 200mm B= 182mm
Mixing Time AVG STD DEV
36 30 32 32.7 3.06
STAS_DOWN SHAFT ANGLE WITH THE BOTTOM 45°
Vessel dim 650x252x130 mm
addition point: on liquid surface over the impeller Impeller speed 480 rpm
A= 180mm B= 126mm α =30
Mixing Time AVG STD DEV
50 54 58 54.0 4.00
A
B
A
B
A B α
For the same impeller and rotational speed the mixing time measured in symmetrical configuration was about 42sec. From the data above reported, it is clear that the position and the orientation of the impeller is a very important parameter, in fact in some case (2
nd, 5
t hand 6
t h) the mixing time measured is 15-20% shorter. This parameter will be better analysed in further works.
STAS_DOWN SHAFT ANGLE WITH THE BOTTOM 45°
Vessel dim 650x252x130 mm
addition point: on liquid surface over the impeller Impeller speed 480 rpm
A= 180mm B= 126mm α =30
Mixing Time AVG STD DEV
42 39 40 40.3 1.53
STAS_DOWN SHAFT ANGLE WITH THE BOTTOM 45°
Vessel dim 650x252x130 mm
addition point: on liquid surface over the impeller Impeller speed 480 rpm
A= 200mm B= 50mm α =45
Mixing Time AVG STD DEV
30 33 35 32.7 2.52
STAS_DOWN SHAFT ANGLE WITH THE BOTTOM 45°
Vessel dim 650x252x130 mm
addition point: on liquid surface over the impeller Impeller speed 480 rpm
A= 200mm B= 50mm α =45
Mixing Time AVG STD DEV
38 38 37 37.7 0.58
A B
α
A
B α
A B
α
4.3 Fluid Flow Pattern
4.3.1 Particle Image Velocimetry 2D
The first issue studied with this equipment, was the full understanding of the flows produced by the impellers. It is very important to have a deep insight into the type of flow generated, also because afterwards is possible to explain results better, like the mixing time. The two impeller analysed with this technique are the 3PBT30 and the STAS DOWN. To study only the liquid flow the two impellers were moved away from the bottom of the vessel to avoid the wall effect. The lasersheet has been set parallel to the long side of the vessel, passing in the middle of the impeller, while the camera on one side. This configuration allowed to catch the images like a cross-section of the
Figure 4-18: Flow pattern for 3PBT30 impeller at 348rpm
impeller.
In Figure 4-18 and Figure 4-19 are shown the vector flow fields of 3PBT30 and STAS DOWN impellers respectively; both conditions are for ε
T=1 W/m
3, meaning that the 3PBT30 rotates at 348rpm, while the STAS DOWN at 303rpm. During acquisition, the maximum velocities measured, v
m a x, for the two cases were 0.37 and 0.56m/sec respectively, while the hypothetical maximum velocity should be the velocity of the blades, v
t i p, which is 0.87 for 3PBT30 and 0.62 m/sec for STAS DOWN. Comparing the v
t i pwith the real velocities developed, it can be notice that the STAS impeller is able to transfer higher speed to the liquid despite a lower tip speed. However, looking at the field vector colours, the pumping capacity of the 3PBT30 is greater than the STAS impeller; in fact the yellow and the green areas are larger.
Figure 4-19: Flow pattern for STAS DOWN impeller at 303rpm
The last aspect could explain the reason why the mixing time for the STAS DOWN is longer than the 3PBT30 impeller.
To better understand the flow pattern developed by the impellers inside the vessel, two other conditions have been analysed and shown in Figure 4-20 and Figure 4-21. The first image is referred to the
Figure 4-20: Flow pattern for 3PBT30 in standard positionFigure 4-21: Flow pattern for 3PBT30 with the angled plane
standard working position, while the second one is the position used during the initial power measures. As shown, the plane angled at 45°
and positioned underneath the impeller, is a good solution to approach that problem, because the flow produced in these positions is very similar , even though the case with the plane shows a higher maximum velocity.
From Figure 4-18 is possible to estimate the flow number, f , defined as above:
Where Q is the impeller pumping flow rate, N the impeller speed [rps], and D the diameter of the 3PBT30. In fact, Q can be calculated taking a mean velocity of liquid multiplied for a corona circolare areas; the expression of f becomes:
From the PIV results v can be assumed equal to 0.2 m/sec, while concerning the inner diameter of the corona circolare, d , the shaft diameter was chosen, 10mm. Substituting these values in Eq. 4-2 the flow number assumes the value of 0.55; this value is completely in accordance with the results of the formula suggested by Nienow (1998)
17, which provides the flow number as a function of Po :
In fact, using Eq. 4-3 the flow number assumes the value of 0.57.
D
3N f Q
= ⋅
Eq. 4-1( )
3 2
2
/ 4
D N
d D f v
⋅
⋅
−
= ⋅ π
Eq. 4-23 /
76
1.
0 Po
f = ⋅
Eq. 4-3Also for the STAS DOWN impeller, the two different positions were analysed in Figure 4-22 and Figure 4-23. This case is even better the previous one, in fact the two flows are almost identical and the maximum velocity magnitude is the same for both of them.
Figure 4-22: Flow pattern for STAS DOWN in standard position
Figure 4-23: Flow pattern for STAS DOWN with the angled plane
As mentioned above, comparing Figure 4-20 with Figure 4-22 it can be easily noticed that the 3PBT30 impeller creates a stronger flow, hence the pumping capacity is higher.
4.3.2 Particle Image Velocimetry 3D
After having analysed the 2D liquid flow in proximity of the impellers, the whole vessel has been studied; the aim is to define the main flows inside the reactor and to compare different solutions of impellers with the simple lance. For this purpose the 3D PIV has been used. To calibrate the 2D PIV system is quite simple, because it is only necessary to find the ratio between the real dimensions observed and the number of pixels of the image recorded. On the contrary the 3D PIV calibration requires long procedure and it needs meticulousness and high accuracy. In fact, TSI Incorporated has provided, with the rest of the equipment, a special target which has to be aligned with the lasersheet in the position of interest; after that, the two cameras record two different images and the calibration software is able to create a file of calibration using the marks printed on the planes of the target.
Once the calibration was concluded, the vessel was divided in five
sections, and in each of these an average stereoscopic flow field was
elaborated on 15 pairs of images for each camera. In Figure 4-24 is
shown the equipment during the acquisition in one section. After that
all the sections were assembled in one graph to show the
development of the liquid flow along the whole vessel, as shown in
the following graphs.
Two different resolutions were used, in fact to compare the different systems a low resolution analysis were done, while a high resolution offers an insight into the flow developed inside the vessel. In this kind of representation the orientation of the vectors is the real direction of the flow, their sizes proportional to the velocity magnitude , while the colour shows the value of the property listed in the legend.
To compare the different configurations the same specific energy
dissipation rate, ε
T=1 W/m
3, was used for all of them. Four
configurations have been examined: lance with a gas flow rate of 1.0
l/min, 3PBT30 rotating at 348rpm, STAS DOWN impeller at 303rpm and
a combined apparatus with the lance and the STAS impeller together
Figure 4-24: Stereoscopic PIV equipment during an experiment(Figure 4-25, Figure 4-26, Figure 4-27 and Figure 4-28 respectively). For the impeller configurations the five sections are positioned at 250, 330, 410, 490 and 555 mm from the impeller entry side, while for the single lance system it was possible to analyse a section more at 150 mm. This last measure is impossible to catch with the impeller position, because they are black painted, so the light can not cross the material and a shade covers the region of interest. Furthermore using a lower resolution the very bottom part of the tank is missing (about 40mm); if this is not a problem for the lance system, it is an important absence for impellers configuration.
In Figure 4-25 is shown the flow field generated by introduction of air using the lance, the maximum velocity measured during the acquisitions was 0.04 m/s on the surface above the injection point.
Furthermore looking at the right side of the vessel, all the vector are very short, meaning that in this region the liquid is almost steady in accordance what was noticed during mixing time experimentations.
The section at Z=250mm, shows how the liquid follows the rising bubble (injection point at 200mm), reaching the surface and then changing direction from vertical to horizontal.
For 3PBT30 and STAS DOWN impellers, as mentioned above, is useless
to look at the values of the maximum velocity; however the Figure
4-26 and the Figure 4-27 are helpful in indicating the magnitude of
the velocity field along the vessel. In addition in Figure 4-27 is
important to notice that a sort of vortex is generated on the right part
of the vessel, far from the impeller. In fact, looking at the horizontal
component of the vectors in front of the graph (X˜ -100mm) and in the
back (X˜100mm), they have opposite directions; this was noticed also
during the mixing measures, looking from above of the vessel a vortex rotating in clockwise direction appeared. This phenomenon is more evident in Figure 4-32.
Figure 4-28 is concerning a combined system, e.g. STAS DOWN
rotating at 303rpm and lance with a gas flow rate of 1.0 l/min. Also in
this case the maximum velocity is not so relevant, because the
bottom part is missing; however is clear that the main flow is
generated from the impeller, despite a strong contribute given by the
rising bubbles. As in the previous case, the STAS impeller produces
the same vortex, also with the presence of the lance.
Figure 4-25: The whole flow field vector of
lance system Air 1.0 l/min Figure 4-26: The whole flow field vector of 3PBT30 impeller system, 348rpm
Figure 4-27: The whole flow field vector of
STAS DOWN impeller system, 303rpm Figure 4-28: The whole flow field vector of combined system, STAS DOWN 303rpm + Air 1.0 l/min
In the following figures are shown the results of 3D PIV concerning the 3PBT30 and STAS DOWN impellers at high resolution. Velocity magnitude and horizontal component (Z direction) are shown with this technique. On the contrary of the low resolution, in this case it was possible measure the velocity flow field till the very bottom of the vessel. As shown in Figure 4-29 and in Figure 4-30 the flows and maximum velocities generated from the two impellers are substantially different in proximity of them; in fact the section at 250mm (only 50mm far from the impeller) shows different main direction and module of velocity for the two impellers, despite both of them are defined as axial impeller and the ε
Tis the same.
Concerning the flow direction, in Figure 4-32 is possible to understand better the vortex created in the right side of the vessel and the reason of its presence. In fact, comparing the flow generated by the STAS with the one created by 3PBT30, Figure 4-31, is clear how the first one is more direct to the back wall than the other; this flow with the wall effect produce that sort of vortex observed in all the mixing time experiments. After these results the necessity to investigate some asymmetrical configurations in terms of mixing time came out.
Finally, as already mentioned in section 4.3.1, the 3PBT30 has a
greater pumping capacity than the STAS, so, at this distance from the
position of action, the liquid maximum speed is 0.187m/sec for the
3PBT30 and 0.142m/sec for the STAS, in accordance with the values
found in 2D PIV results. The last two figures show the horizontal
component in Z direction of the velocity, instead of the velocity
magnitude.
Figure 4-29: 3D velocity field for 3PBT30 impeller at 348rpm, Velocity Magnitude
Figure 4-30: 3D velocity field for STAS impeller at 303rpm, Velocity Magnitude
Figure 4-31: 3D velocity field for 3PBT30 impeller at 348rpm, Horizontal Component
Figure 4-32: 3D velocity field for STAS impeller at 303rpm, Horizontal Component
4.4 Bubble Size
The last subject experimentally analysed is the distribution of bubble sizes. This is an important factor in cleaning aluminium processes; in fact, the reaction between chlorine and aluminium is possible only on the contact surface between the two phases. To facilitate the reaction it is preferable to work with a large superficial area. With this aim the equipment described in section Errore. L'origine riferimento
non è stata trovata. has been used to have an insight of bubblesdimensions in the region just above the impeller. U sing this apparatus it was not possible to catch appropriate distributions for all the configurations analysed in the previous sections. In fact, the use of the camera through the microscope is a big limit for the amplitude of the area of interest. At lower magnitude ratio the observable area covered by the frame was 9x12mm, allowing to record the bubbles of smaller size on the frames properly, while recording only a portion of the bigger ones. Furthermore the shape of this kind of bubbles is not perfectly spherical, so it is not correct to estimate their diameter looking at only a portion of them. Another problem found during the experimentation is about the short depth of field of the optical system (e.g. the range of the vessel on focus); in fact for a bubble out of focus is hard to define the borders to calculate the diameter
14,18.
Figure 4-33: Examples difficult bubble frames: no spherical, only a portion, out of focus.
About the difficulty to determine the bubble diameter, it has to be noticed also that in frames where bubbles are overlapping is not easy to understand the boundary between the two bubble.
After this introduction, only for low gas flow rate (0.1 l/min) has been possible to find appropriate distribution in two different impeller regimes, ε
Tequal to 2 and 7 W/m
3, shown in Figure 4-34 and Figure 4-35 respectively.
The first graph in Figure 4-34 is the bubble distribution when no impellers are acting, but only the lance is generating bubbles inside the vessel. The distribution is very narrow and the mean diameter is about 3-3.5mm; at 438rpm ( ε
T=2W/m
3) it seems that the 3PBT30 does not have any effect on gas spreading, in fact the mean diameter does not change. On the other hand the STAS impellers show a very similar behaviour in terms of gas spreading, generating a quite wide distribution around the mean value of 2-2.5 mm. For STAS impeller it seems that the direction of pumping does not influence the capacity of gas dispersion, probably because the type of cavities formed behind the blades is the same.
Increasing the power input in the system to 7W/m
3, better
performances are obtained for all the impellers. In fact, as shown in
Figure 4-35 the mean diameter is about 1mm. Also in this
configuration the STAS impellers are working better than the 3PBT30, in
fact the distribution for these impellers is narrower, even if the mean
value is the same. Comparing the behaviour of the single impellers
between the two trials, only the 3PBT30 shows a big efficiency
improvement when increasing impeller speed. This is due to the
number of blades; in fact a bigger number of blades increases gas
spreading capacities.
Graph Only gas 0.1 l/min
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 10 20 30 40 50 60
Graph 3PBT30 0.1-438
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 10 20 30 40 50 60
Graph STAS DOWN 0.1-380
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 5 10 15 20 25 30
Graph STAS UP 0.1-380
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 5 10 15 20 25
Figure 4-34: Bubbles size distributions with gas flow rate of 0.1 l/min and
ε
T=2 W/m3Graph Only gas 0.1 l/min
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 10 20 30 40 50 60
Graph 3PBT30 0.1-750
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 5 10 15 20 25 30 35
Graph STAS DOWN 0.1-650
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 10 20 30 40 50 60
Graph STAS UP 0.1-650
size (mm)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
percentage (%)
0 10 20 30 40 50
Figure 4-35: Bubbles size distributions with gas flow rate of 0.1l/min and
ε
T=7 W/m3An important parameter to evaluate the efficiency of the spreading systems is the Sauter Mean Diameter ( d
3 2); in fact if the distributions give a good insight in how the impellers are able to interact with the gas, the d
3 2is the equivalent surface area mean diameter
13,19. It is calculated as:
This is the diameter of a sphere having an area equivalent to the mean of the areas of all the bubbles having the diameters measured during each experiment.
Table 4-2: Sauter Mean Diameter for impeller systems
Configuration
d
3 2[mm]
Air Lance 0.1 l/min 3.28
3PBT30 2 W/m
33.32
STAS DOWN 2 W/m
32.73
STAS UP 2 W/m
32.59
3PBT30 7 W/m
32.67
STAS DOWN 7 W/m
31.76
STAS UP 7 W/m
31.40
As shown in Table 4-2, the Sauter mean diameter confirms clearly that STAS impellers are better than the 3PBT30, and how the upward pumping is the best of two.
∑ ∑
=
23
32