You are on page 1of 21

Packed Beds: Pressure Drop versus

Fluid Velocity and the Ergun


Equation

Group 10

Paul Chung
Rick Koontz
Beth Newton

February 14, 2002

Group 10 Packed Beds Page 0


Table of Contents

Abstract .2

Introduction ...3

Theory 3

Equipment and Procedure ...6

Results and Discussion .7

Error Analysis9

Future Experimentation .10

Nomenclature ..12

References ...13

Appendix Index ..14

A. Sample Calculations ...A1 (15)

B. Calibrations..B1 (16)

C. Data ...C1 (17)

Group 10 Packed Beds Page 1


Abstract:

Erguns equation is the relationship between fluid velocity, the type of packing,
and the pressure drop over the distance of a packed column. We found that the behavior
of the pressure drop followed the trend of Erguns equation very well in smaller columns.
Our results also indicated a consistent pattern in fluidization behavior. Many factors of
error did incur due to the lack of time to collect, and in continuing experiments, we
suggest collecting more data in the pre-fluidization range, changing the packing material,
and fixing the orientation of the packing.

Group 10 Packed Beds Page 2


Introduction:

Packed beds are used in the chemical industry for many purposes including
catalytic reactors, gas-liquid adsorption, absorption, as well as many other uses
(Geankoplis 118). They usually consist of a tower containing small bits of material,
examples including: glass marbles, ceramics, plastics, pea gravel, or mixtures of
materials. There are inlet and outlet pores for the fluid and two pressure nodes above and
below the packing. The change in pressure can be measured in response to changing
fluid flow at the inlet, different column diameters, different types of packing, and
different fluids.
The main task in this experiment was to measure change in pressure across a
packed column as a function of flow rate. This information is important when designing
the pump for the fluid, in order to decrease costs of the system, but also to maintain the
optimal operating conditions and to maximize the product. Usually a packed tower will
operate at the maximum economical pressure drop (Eckert 56). Our purpose was to test
our data of change in pressure vs. fluid velocity under various conditions with the Ergun
equation, developed by Sabri Ergun in late 1940s/early 1950s (Ergun 89). Erguns
model takes into account both viscous and kinetic energy changes, as a result of the fluid
and packing properties, and can be used over a wide range of flows. The variables we
tested were fluid velocity, column diameter (6 inch and 3.5 inch), and particle type (1/2
inch glass marbles and pea gravel). This task led to other subtasks including calibrating
the rotameters for fluid flow, calibrating the manometer for pressure drop, and
determining the void fraction of both the glass spheres and the pea gravel.

Background and Theory:

In the design of a packed column, many parameters need to be taken into account
in order to control the system. As Eckert mentions, the pressure drop should be at a
maximum in order to produce the best product, but the cost of the pump for the fluid is
the trade-off (Eckert 56). So in order to maximize the pressure drop, but minimize the
power required of the pump, Ergun suggests considering these factors: (Ergun 89)
(1) Rate of fluid flow

Group 10 Packed Beds Page 3


(2) Viscosity and density of fluid
(3) Closeness and orientation of packing
(4) Size, shape and surface of particles

In earlier times it was observed that pressure drop was proportional to velocity at
low flow rates, and proportional to the square of the velocity at high flow rates. Osborne
Reynolds was the first to formulate the resistance by friction on the motion of fluid as the
sum of these two conditions:

P 2
:= aU + b U ( 1)
L

where a and b are representative of packing and fluid properties (Ergun 89)

Determined to figure out the nature of a and b, Carman and Kozeny found that for
viscous flow, the change in pressure was proportional to (1- )2 / 3. It was also
experimentally determined, by way of 640 experiments, that a constant of 150 was also a
factor in the equation. These findings resulted in the Carman-Kozeny equation for
change in pressure under viscous flow: (Ergun 90)

P 150 (1 ) U
2
:= ( 2)
L
3

2
(D p ) 2

At the same time, Burke and Plummer discovered that change in pressure at turbulent
flow, resulting from kinematic energy loss, was proportional to (1- )/ 3. There was also
a constant of 1.75 found to be relevant at the turbulent flow, resulting in the Burke-
Plummer equation for change in pressure at turbulent flow: (Ergun 90)

1.75 ( 1 ) U
2
P
:= ( 3)
L 3
Dp

Group 10 Packed Beds Page 4


It was Ergun and Orning that put these two equations together, and through much
experimentation found that it was accurate for a wide range of Reynolds numbers.

150 U ( 1 ) L 1.75 U ( 1 ) L
2 2
( 4)
P := +
2 2 3 3
Dp Dp

As Ergun stated previously, particle size, shape, and closeness are also important in his
equation. To account for closeness and position of the particles, the void fraction needs
to be calculated, and to account for the size and shape of the particles, the effective
particle diameter (Dp) and the sphericity() of the particles also need to be calculated.
The Ergun equation was designed for fluid flow up until the fluidization point,
when the pressure drop multiplied by the cross-sectional area equals the gravitational
force, and the particles start to move:
P
L
( )(
:= 1 mf p g )
mf
(5)
where mf, and Lmf, are the void fraction and length of packing at minimum
fluidization and p is the particle density.

Also using velocity and Reynolds number at fluidization, Umf and Nremf, the Ergun
equation (4) at the fluidization point can be written:

N
NRemf
0 := 150 ( 1 mf ) + 1.75
( Remf ) 2
D3 ( ) g
2 ( ) 3 3
p 2
mf ( mf )
(6)
Past this point the Ergun equation does not hold. It was our goal in this experiment to
test the accuracy of this equation pre-fluidization, and to see to what degree the Ergun
equation does not fit for velocities past fluidization.

Equipment and Procedure:

Our equipment consisted of a plexi-glass column with an inlet and an outlet for
water. Water flowed in the bottom of the unit and out the top. The column was
graduated from 0 to 36 on one side. At the top and bottom were connections for digital
manometer hoses. The flow was controlled by a series of rotameters, each with a

Group 10 Packed Beds Page 5


different gauge. Finally, the column had a valve in the bottom for a drain, once the
experiment was completed.
Before our procedure could start, we needed to check and calibrate our
equipment. First, a rotameter was spot checked against previous calibrations to obtain a
relationship between a given meter flow reading and the actual volumetric flow rate
(manual, 1). An empty barrel was filled with water to the first accurately measurable
graduation. Then, the rotameter was turned to a set flow reading. The barrel was filled to
the next graduation and the time it took to reach that graduation was timed with a
stopwatch. This experiment was repeated for a second rotameter reading. The flow rates
were calculated for the given rotameter reading and compared to the previous results.
Next, the digital manometer was calibrated. To do so, we filled the column of
water to the first graduation mark (36). The manometer was tared to a reading of 0
units. Water was allowed to drain from the column and the change in height and the
pressure drop were recorded. The manometer was calibrated in inches H2O so the
pressure drop should have equaled the change in height of the water column in inches.
Third we needed to calculate the void fraction of our packing material in the
packed state. To do this, we filled a 4 L graduated cylinder with pea gravel. The gravel
was packed the same as in our experiment column. Then, water was added until the level
of water was even with the level of the gravel. The amount of water used to get to this
point was recorder as was the total volume of the mixture. The void fraction was:

Volume
water
Void := ( 7)
Volume
mixture

Finally, we could move to our actual procedure. Our procedure attempted to


measure the pressure drop across a packed bed. Using the apparatus described above, we
filled the column approximately half way, to a depth of 18 with pea gravel. The gravel
was then packed in the column by lightly tapping the side of the column with a metal rod.
This allowed the gravel to completely settle into a most packed state.
With the column packed, the flow was gradually increased through the range of
the rotameter. Date points were measured in even increments depending on the rotameter

Group 10 Packed Beds Page 6


used. Once the upper limit of the meter was reached, we slowly lowered the flow
through the range of the meter. Data points were taken for the pressure drop across the
column and the height of the packing material as a function of the flow. Finally,
measurements were taken in random rotameter increments so determine if there was any
difference between increasing or decreasing each increment.
After data were taken for a given packing material and column diameter
combination, either the material or the column was switched. For our experiment we
used pea gravel and glass marbles for our materials and 3.5 and 6 diameters for our
columns. Our procedure was repeated for all column diameters and materials.

Results and Discussion:

The data we collected with the pea gravel produced values that were consistent
for the 3.5 inch column to Erguns prediction and inconsistent with the 6 inch column to
Erguns prediction. The pressure drop values in pounds-force/sq feet were compared as
in Figure 1.

Pea Gravel (3.5 inch column)

700
600
deltaP (lbf/ft^2)

500 Ergun
400 Ascending Velocity
300 Descending Velocity
200 Random Velocity
100
0
0 0.05 0.1 0.15 0.2 0.25
v (ft/s)

Figure 1. Pressure drop versus velocity for Pea Gravel with fluidization. The theoretical plot of Ergun
involves the void fraction(.32), sphericity(.7), and effective particle diameter(.01 in). The runs were
executed to observe any dependence upon accelerating or decelerating the fluid velocity. A random run
was also done to further disprove this dependence. The bed fluidized consistently at the same velocity, v.
The Ergun equation was not designed to predict after fluidization occurs, which our data corresponds.

Group 10 Packed Beds Page 7


Pea Gravel (6 inch column)

600
500

deltaP (lbf/ft^3)
400 Ergun
300
Ascending Run
200
100
0
0 0.05 0.1 0.15 0.2
v (ft/s)

Figure 2. Pressure drop versus velocity for Pea Gravel with fluidization. The theoretical curve is
produced by Erguns equation, the same as from Figure 1. A larger column was used this run, producing
values that were initially larger than what Ergun predicts. Again, at the point of fluidization, the curve
drops well below the theoretical curve.

The half inch glass marbles produced similar results as the pea gravel. In the smaller
column, the values of the pressure drops from multiple runs coincided with Erguns
prediction, shown in Figure 3. The effect was better observed since the velocity o f
fluidization is much higher for glass marbles than for pea gravel. Again in the 6 inch
column, the observed values are consistently poor.

1/2 in. Marbles (in 3.5 inch column)

400
350
300
deltaP (lbf/ft^2)

250 Ascending Velocity


200
Ergun
150
100 Descending Velocity
50
0
-50 0 0.2 0.4 0.6
v (ft/s)

Figure 3. Pressure drop versus velocity for glass marbles with fluidization. The Ergun predicted values
were based off of the same factors as Figures 1 and 2. The sphericity is assumed ideal ( = 1); the void
fraction calculated to be .4; and the effective diameter is .5 inches. The experimental data for the marbles
were consistently better than the pea gravel. The data again deviates from the predicted values at the
fluidization point.

Group 10 Packed Beds Page 8


1/2 in. Marbles (6 inch column)

200

deltaP (lbf/ft^3)
150

100 Ergun

Ascending
50 Velocity

0
0 0.05 0.1 0.15 0.2
v (ft/s)

Figure 4. Pressure drop versus velocity for glass marbles with fluidization. The Ergun predicted values
were based off of the same factors as Figure 3. The experimental data for the marbles were consistently
poor like the pea gravel in the larger diameter column. The data does not correspond with the theoretical
values at all. The values collected were much greater than the Ergun theoretical values. Another souce of
error must be occurring with the 6 inch column.

Error Analysis:

Just by visual observation of the plotted experimental data versus Erguns


theoretical values, large amounts of error can be observed due to sphericity measurement,
effective diameter measurement, calibrating the rotameters, and most importantly
calculating void fraction. All of these possiblities of error are calculated constants and
variables in the Ergun Equation. Sphericity was not calculated, but was taken as an
average of the values for sand and broken glass. This was an educated recommendation,
but it could have been incorrect. The proper method would have been to calculate the
effective diameter, which we took from a past group's data, which is used to calculate the
sphericity. The effective diameter taken from the previous group may be inaccurate due
to reduction in diameter by friction over time. Since both the pea gravel and glass marble
trials with the 6 inch column had enormous error, there could possibly be a defect in
connections for the manometer causing an increase in the pressure drop. The most
significant source of error is calculating the void fraction of the packing material. No
matter what type of packing is used for the trials, there will always be a minimum of 32%
void fraction. The total void fraction will vary due to packing effects. The marbles are
usually hexagonally face centered but can also be cubic-face centered, creating a larger
void fraction than expected. We also used a graduated cylinder that wasn't the size of the

Group 10 Packed Beds Page 9


column, so we could not duplicate the exact effects in the experimental column. When
analyzing the Ergun equation, one will find that a small change in void fraction will incur
a much larger change in the pressure drop. Other unexplainable factors were air bubbles,
which were a reoccurring problem for the calibration of the rotameters and the
calculation for the void fraction of the packing material. Following the procedure to
measure void fraction from past groups proved to be unsuccessful, since air bubble
removal was a constant factor. We utilized the rotameter calibrations from previous
groups. Even though we spot-checked the flowrates, a little error is also present when the
graduated tank is in large units (5 liter increments). The rotameters fluctuated a large
amount due to uneven flow of water from building's water supply. We had to take an
average of the fluctuations and adjusted the rotameter through trials to keep constant flow
rate.

Future Experimentation:

In running this experiment we thought our goal was to determine the point of
fluidization and have data reflect this point. Also, we thought data beyond the
fluidization point was just as necessary as data before fluidization occurred. Upon
analysis of our data versus Erguns equation, we found that our data points after
fluidization occurred we not as significant as those before fluidization. Ergun accurately
predicts the pressure drop across a packed bed when there is no fluidization. However,
once fluidization occurs, the Ergun equation becomes very inaccurate and
misrepresentative of the actual data. So, one point of future inquiry would be to find
more data in the pre-fluidization range. Then, we could compare more closely our data to
the data Ergun predicts.
Another point of future experimentation would be the packing material itself.
While we varied our material, we only used 2 different materials. And, we did not mix
materials any within our experiment. In consulting our references, we found that the
Ergun equation was especially accurate in predicting the pressure drop across a bed
packed with Raschig Rings. We think it would be interesting to work with these rings to
see just how accurate the Ergun model is in that case. Similarly, we would like to use

Group 10 Packed Beds Page 10


mixed packing material for our packed bed. If we were to mix sand or another fine
particle in with one of our more coarse particles, we could decrease the void fraction
considerably. We would like to experiment with the void fraction and the accuracy of
Erguns equation.
Finally the packing orientation could be another area of further study. First we
would like to fix the bed height through the use of a rigid rod with a retainer for the
packing material. This support would be capable of allowing water to flow through the
retainer, while keeping the bed height to a constant level. While we surmise it would, we
think it would be beneficial to learn whether the Ergun equation would continue to
predict the pressure drop across a bed that had a constant height. Lastly, we would like to
fix the orientation of the packing in a prearranged pattern. We noticed the marbles
tended to pack almost in a close-packed hexagonal arrangement. This arrangement is the
densest packing for a set of spheres. We would like to try a face centered cubic pattern
and see if the variance of void fraction affects the Ergun model.

Group 10 Packed Beds Page 11


Nomenclature

Uppercase:
D Inner diameter of column [=] ft
Dp Effective particle diameter [=] ft
g Gravitational acceleration = 32.174 ft-lb/lbf -s2
L Height of packing [=] ft
Lmf Height of packing at point of minimum fluidization [=] ft
NRemf Reynolds number at point of minimum fluidization = Dvmf / [=] unitless
U Fluid velocity [=] ft/s
Umf Fluid velocity at point of minimum fluidization [=] ft/s

Lowercase:
a Representation of packing and fluid characteristics at laminar flow
b Representation of packing and fluid characteristics at turbulent flow

Greek:
P Pressure drop across packing [=] lbf/ft2
Void fraction of packing [=] unitless
mf Void fraction of packing at point of minimum fluidization [=] unitless
Viscosity of fluid [=] lb/ft-s
Sphericity (surface area of sphere/ surface area of particle) [=] unitless
Density of fluid [=] lb/ft3
p Density of packing [=] lb/ft3

Group 10 Packed Beds Page 12


References

Eckert, John S., Design Techniques for Sizing Packed Towers. Chemical Engineering
Progress, Vol. 57, No. 9 (1961):54-58.

Ergun, Sabri, Fluid Flow Through Packed Columns. Chemical Engineering Progress,
Vol. 48, No.2 (1952):89-94.

Flow Through Packed Beds, Experiment Info, Packed Beds, Resources, Manuals,
Manual - PDF, http://rothfus.cheme.cmu.edu/tlab, 2002.

Geankoplis, Christie J., Transport Processes and Unit Operations. 3rd ed., New Jersey:
Prentice Hall, 1993.

Group 10 Packed Beds Page 13


Appendix Index
B. Sample Calculations ...A1 (15)

B. Calibrations..B1 (16)

C. Data ...C1 (17)

Group 10 Packed Beds Page 14


Appendix A:

Sample Calculations (for Marbles in 3.5" Column)


Unit Conversions and Constants

4 lb ft lb
Dcyl := 3.5 in := 1 Dp := .5 in := 8.77 10 gc := 32.174
ft s lbf s
2

lb lbf
:= 62.3 r := D p .5 L := 1.54 ft 1inH2O := .03613
3 2
ft s

Calculation of Void Fraction


Vtot := 1525 mL VH2O:= 605 mL

VH2O
:= = 0.397
Vtot

Calculation of Volumetric Flowrates from Rotameter Readings (W-3)


W := 10
mL mL mL
F := 48.785 W + 32.644 F = 520.494
s s s

Calculation of Fluid Velocity

mL
F := 520.494
s
2
A :=
Dcyl F ft
v1 := v1 = 0.275
2 A s

Theoretical Value of Pressure Drop using Ergun's Equation

150 v ( 1 ) L 1.75 v ( 1 ) L
2 2
ErgunP( v ) := +
2 2 3 3
gc Dp gc Dp

lbf
ErgunP( v1) = 97.402
2
ft

A1

Group 10 Packed Beds Page 15


Appendix B:

W-3 gage

reading: 4 4 18 18
initial volume (gal) 10.0 17.5 30.0 37.5
final volume (gal) 12.5 20.0 35.0 42.5
time (sec) 40.000 43.000 22.000 20.000
change in volume (mL) 9463.500 9463.500 18927.000 18927.000
flow rate (mL/sec) 236.59 220.08 860.32 946.35

W-2 gage

reading: 20 20 20 70 70
initial volume (mL) 10000.0 30000.0 50000.0 100000 120000
final volume (mL) 20000.0 40000.0 60000.0 110000 130000
time (sec) 120.000 110.000 110.000 34 35
flow rate (mL/sec) 83.33 90.91 90.91 294.12 285.71

W-3 W-2 y = 4.0306x + 7.771


y = 48.214x + 35.477 R2 = 0.9985
2
R = 0.9916
1000.00 350.00
300.00
flowrate (mL/sec)

800.00
flowrate (mL/s)

250.00
600.00 200.00
400.00 150.00
100.00
200.00
50.00
0.00
0.00
0 5 10 15 20
0 20 40 60 80
reading
reading

SPOT-CHECK for Calibrations for Rotameters (Relationships taken


W-2 W-3 from Group 8
y=4.0362*x+3.2292 *** y=48.785*x+32.644 *** April 1998
20 84 mL/s 4 227.78mL/s Boyars, Conley, Didier)
70 285.76 mL/s 18 910.77mL/s

B1

Group 10 Packed Beds Page 16


Appendix C:

Raw Data:

Pea Gravel, small column

(W-2 Rotameter
Readings) Ascending
Water flowrate Pressure Drop (in. H2O) Pea Gravel Ht (in)
0 0 15
15 9.8 15
20 15.8 15
25 17 15.5
30 18.1 15.75
35 16.5 16.5
40 16 17
45 16 17.5
50 16 17.75
55 15.9 18
60 16.3 18.5
65 16.5 19.25
70 16.7 19.75
75 16.9 20.5
80 17 21
85 17.2 22
90 17.2 23

(W-2 Rotameter
Readings) Descending
Pressure Drop
Water flowrate (in. H2O) Pea Gravel Ht (in)
85 17.2 22.75
80 17.1 22
75 17.1 21.5
70 17 20.5
65 16.8 20
60 16.6 19.5
55 16.5 19
50 16.6 18.5
45 16.6 18
40 16.6 17.5
35 16 17
30 14 16.75
25 12.2 16.5
20 9.6 16.5
15 7.2 16.5
10 4.6 16.5
C1

Group 10 Packed Beds Page 17


(W-2 Rotameter Readings) random
Water flowrate Pressure Drop (in. H2O) Pea Gravel Ht (in)
10 3.5 17
15 5.3 17
20 6.6 17.25
25 9.6 17
30 16.2 17
35 15 17
40 15.6 17.5
45 16.4 18
45 16.8 18
50 16.6 18.5
55 16.8 19
60 16.6 19.5
65 17 20
65 17.1 20
70 17.2 20.5
80 17.2 22.5
85 17.2 22.5

Pea Gravel, large column

(W-2 Rotameter Readings) Only one run


Water flowrate Pressure Drop (in. H2O) Pea Gravel Ht (in)
1 8 13.75
2 12.4 13.75
3 16 13.75
4 15.5 14.25
5 14.5 14.5
6 14.5 15
7 14.7 15.25
8 14.9 15.75
9 15.2 16
10 15.2 16
11 15.2 16.5
12 16.2 17
13 16 17.25
14 16.5 18
15 17 18.5
16 16.5 19
17 16.6 19.5
18 16.4 20.5
19 16.2 20.75
20 16.3 21

C2

Group 10 Packed Beds Page 18


1/2 inch Marbles, small column

(W-2 Rotameter Readings) increasing flow rate


meter reading Pressure Drop (in. H2O) Marble Ht (in)
0 0 18.5
1 0.3 18.5
2 0.9 18.5
3 1.8 18.5
4 2.8 18.5
5 3.7 18.5
6 5 18.5
7 6.6 18.5
8 7.6 18.5
9 9 18.5
10 11.2 18.5
11 13.7 18.5
12 15.3 18.5
13 16.3 19
14 16.8 19.75
15 17.2 21
16 17.2 22
17 18.2 22.5
18 18.3 24.5
19 18.6 25.5
20 18.6 26.5

(W-2 Rotameter Readings) decreasing flow rate


meter reading Pressure Drop (in H2O) Marble Ht (in)
0 -0.1 19
1 0.2 19
2 0.9 19
3 1.7 19
4 3.8 19
5 3.8 19
6 5 19
7 6.6 19
8 7.7 19
9 10.1 19
10 11.4 19
11 13.6 19
12 16 19
13 16.7 19.25
14 17 20
15 17.1 21
16 17.4 22
17 17.8 23.5
18 18 24.5
19 18.2 25

C3

Group 10 Packed Beds Page 19


1/2 inch Marbles, large column

(W-2 Rotameter Readings) Increasing flowrate


meter reading Pressure Drop (in. H2O) Marble Ht (in)
0 0 18.5
1 0.6 18.5
2 1.6 18.5
3 3.4 18.5
4 3.9 18.5
5 4.3 18.5
6 5.4 18.5
7 6.1 18.5
8 7 18.5
9 8.4 18.5
10 9.5 18.5
11 11.1 18.5
12 12.6 18.5
13 14.4 18.5
14 16.5 18.5
15 18.5 18.5
16 20.6 18.5
17 23.4 18.5
18 25.7 18.5
19 28.2 18.5
20 32.1 18.5

C4

Group 10 Packed Beds Page 20

You might also like