HIGH VOLUME CENTRIFUGAL OIL-WATER SEPARATION

Dave Meikrantz
Charles Schardin
Lawrence Macaluso
Alfred Federici
H. William Sams, I11
CINC
1 125 Business Center Circle
Newbury Park, CA 91320 USA
ABSTRACT
The recovery of spilled oil and hydrocarbon fuels
into marine waters is often limited by the lack of
rapid oil-water separation capability. Therefore, a
new design of liquid-liquid centrifuge has been
developed and tested at flow rates of 2-200 gallons
per minute. Separation efficiencies of greater than
99% have been achieved over a wide range of oil
viscosity and oil-water ratios. Operational
equilibrium is rapid and is maintained during changes
of oil-water ratios, flow rates, and flow interruptions.
The separator appears unaffected by normal
shipboard motion from waves. Laboratory tests at
fixed angles of 35 tilt did not inhibit separation
performance. These features coupled with compact
size and relatively low weight versus throughput
make this technology an attractive candidate for
improving marine oil spill recovery. Separation
efficiency data on tests of water versus No. 2 diesel,
crude oil, and I S0 460 gear oil will be presented for
three sizes of separators.
INTRODUCTION
Ocean spills of crude oil and refined petroleum
products are a continuing environmental concern.
Regulations are being implemented in hopes of
reducing the number and size of petroleum spills.
Double hulled tankers, tanker escorts, and numerous
operational rules have been implemented or
considered to varying degrees, but as yet no uniform
global code exists. However, as long as oil products
are transported by tanker or pipeline across open
waters, the potential for spills exist. Therefore, better
methods of spill recovery must be developed.
Following the Ashland diesel oil spill in 1988 and the
Exxon Valdez incident in 1989, the U.S.
Environmental Protection Agency reported
shortcomings of spill control equipment previously
developed and of our ability to respond effectively.
The average recovery rate of spilled crude oil ranges
from 10% to 30%, but recovery of refined fuels and
solvents is even lower. This is due to the rapid
evaporation rate of such products, and the fact that
they leave little or no residue once vaporized.
There are many factors which complicate the
recovery of spilled oil. The size of the spill, the
nature of the spilled oil, response time, location,
weather, personnel and equipment readiness, and
regulations all impact effective spill recovery. The
Exxon Valdez spill forced a re-examination of many
of these issues and todays oil spill response
capabilities have been improved. Several unified
response organizations were formed to select and
stage equipment, provide and coordinate teams for
large spill response, and train commercial fishermen
and boatmen for additional manpower. Contracts
with the general marine industry were established to
help clarify and streamline response to large spills.
Heightened awareness has inspired efforts to improve
oil spill recovery equipment. The search for a high
volume, efficient oil-water separator has been an
important outcome of this process.
No single technology approach will solve the
problems of recovery of every possible petroleum
spill. Viscosities can range from that of water (- 1
centipoise) to that of crude oils, tars, and asphalts (-
2000 centipoise). Weather and time constantly
change the nature of spilled crude oil, making
equipment which worked well on fresh oil ineffective
within a few days. The evaporation rate of the more
volatile components of crude oil directly affects its
21
viscosity. In addition, the loss of these volatile
components changes the ratio of asphaltenes in crude
oil making them much more prone to oil-water
emulsion formation.' Formation of emulsions causes
additional problems arising from increasing viscosity,
substantially hindering recovery operations such as
pumping and skimming. Emulsions also contain
large quantities of water which greatly increases the
storage requirements for the collected liquid. Wave
motion also serves to enhance emulsion formation.
For these reasons, rapid response time is a crucial
factor in recovery of petroleum spills.
As previously noted, liquid storage capacity is
another major consideration which can directly limit
spill response. Since most high volume skimmers are
at best 20% effcient, much more water is recovered
than oil even in the best of conditions. Without real
time oil-water separators employed during the
skimming process, the volume of liquid which must
be stored is at least five times that of the actual spill.
As the slick thins over time the amount of water
skimmed becomes even greater. In large spills,
recovery often stops prematurely because all
available storage is full. On-line separation is
therefore an important requirement for successful oil
spill recovery. Separator tanks employing gravity
alone (1 g), even when combined with coalescers, are
limited by the natural separation rates of the oil-water
mixture. For this reason, 1 g separators which can
process hundreds of gallons per minute are quite
large and weigh many tons when full. Centrifugal
separators, which employ g-forces in the hundreds or
thousands, are much more attractive candidates for
oil spill recovery in marine applications. However,
none of the existing liquid-liquid centrihge designs
have been successfully employed in oil spill
recovery.
A NEW CENTRIFUGAL OIL-WATER
SEPARATOR DESIGN
The new separator design presented is the result of a
patented technology transfer from the Department of
Energy and an aggressive 18 month development
program at CI NC3 The initial goals were to scale up
and commercially produce separators with capacities
from 2 to 400 gallons per minute (gpm). The
separator is a liquid-liquid centrifuge, which includes
a mixing annulus in the area between the rotor and
the rotor housing. A schematic of the device in
operation is shown in Figure 1. Two liquid phases,
either pre-mixed or separate, are pumped into the
annulus between the housing and the rotor. The
spinning rotor mixes the two phases to form a
uniform dispersion. The dispersed mixture then
flows downward to the bottom plate where it is
directed toward the center of the rotor by radial
vanes. The dispersion enters the rotor and is rapidly
accelerated to rotor speed. Baffles within the rotor
prevent re-mixing and allow separation to occur
under the influence of 200-300 g's. The more dense
phase migrates toward the periphery of the rotor and
displaces the lighter phase
Figure 1
toward the rotor center. Weirs control the rate of
escape for each phase and are thus used to set the
desired interface position between phases. The
separate phases leave the rotor and are collected in
two regions of the upper housing which lead to exit
ports as shown in Figure 1. This design provides an
efficient method for the continuous separation of two
immiscible liquids such as water and oil. The rotor
functions as a pump and fills when rotating. This is
an important feature which allows equilibrium to be
maintained during changes of ratios and flow, or
even jn cases of complete flow interruption. Startup
equilibrium is achieved within 30 seconds. Large
changes in phase ratio are tolerated due to the
dispersion formed in the mixing annulus. Air
pumped from skimmer operations to the separator
does not affect equilibrium as it bypasses the rotor
and exits through the liquid outlets. Shipboard wave
motion is also tolerated due to an efficient design and
the centrifugal form within the rotor. Static testing at
a 35" tilt showed no change in performance. Tests
conducted on a large prototype operating at 135 gpm
were successful in four foot ocean swells.
22
EXPERIMENTAL
Flow
V-5 TESTS
Effluent
Result
Separator tests were conducted using various oils,
and fresh water which was adjusted to 1500 i ; 100
ppmtotal dissolved solids by adding solid NaHCO, .
No. 2 diesel was tested to categorize separation
behavior toward fuel and solvent spill cleanup. I S0
460 gear oil, having a viscosity of 1800 centipoise at
2loC, was used to study heavier oil separation from
water. Two kinds of crude oil, Alaska North SIope
(ANS) and San J oaquin, CA (SJ) were also studied.
Separated eMuents were sampled for analysis by an
outside contract laboratory. Standard methods of
analysis were used; water by EPA 413.l(oil and
grease) and oils by ASTM D-96 (bottom sediment
and water). Three sizes of separators were tested:
the two to six gpm V-5, the ten to thirty gpmV-10,
and the eighty to two hundred gpm V-20. In each
separator model, the number following the V-
designation is the outside diameter of the rotor in
inches. The height of the separating zone for each
model is 2.4 times the diameter. A photograph of the
V-20 electric powered separator is shown in Figure
2. It weighs 3,000 pounds, stands eight feet high and
has a 3x3 platform. The hydraulic powered V-20 is
one foot shorter and weighs 400 pounds less.
Crude
Emulsified
ANS
Figure 2
5 20 3 Water: 1000 ppm
oil
The results of tests with No. 2 diesel and ANS crude
oil separations fromwater are given in Table I. The
V-5 rotor speed was 1700 i 50 revolutions per
minute (rpm) which provides 200 gs of centrifugal
force. Water and oil effluent quality from the
separator for both petroleum products tested is very
high. However, since crude oil emulsions are
commonly encountered, some laboratory generated
ANS crude emulsions were tested. The Tested
I I Phase I Rate I I
Crude Oil: 0.05% water
3 Water: 620 ppmoi l
Oil: 5.8% water
Crude I I I oil: 5.4%water
t Emulsified I 120 I 3 I Water: 1300 ppm
I ANS I I I oil I
Crude I I Oil: 5.8% water
Emulsified I 120 I 3 I Water: 540 ppmoil
I
Oil: 5.4% water
I C961de I
emulsion contained 50% water by volume and a
measured viscosity of 300 centipoise at 21 C. The
emulsion was spilled into a 72 diameter test tank
filled with water. A commercial weir type skimmer
was used to provide feed to the V-5 separator at 3
gallons per minute (gpm). Twenty gallons of test
emulsion were tested in four individual five gallon
batches. All four batches were separated while
adding Exxon BREAXIT OEB-91M emulsion breaker,
added directly to the mixing annulus at the separator
23
feed point. The dosage rate was 200 2 50 ppm.
Samples were taken in the midpoint of each 5 gallon
cleanup test. The water effluent averaged 1000 ppm
oil and the oil effluent averaged 5.7% water. The oil
content in water was approximately twice that of the
previous non-emulsified ANS crude test. The water
content of the recovered oil was almost ten times
better than the input oil-water concentration of 50%.
A third series of tests involving recovery of I S0 460
gear oil were conducted to determine what effect
high viscosity oils had on separation. The viscosity of
this oil at 21 C was 1800 centipoise. Again, five
gallons were spilled on the test tank surface and
recovery was made by skimming at 3 gpm. No
emulsion breaker was used in this study. Five
samples were taken during the recovery of this oil.
The results, in chronological order are given in Table
2.
Table 2
V-5 Separator Test DATA
Input Oil
Concentration
Input Effluent
Flow Rate Result
66% 20 I Water: 1200 ppmoil
Oil: 1.2% water
Oil: 1.0% water
Oil: 1.6% water
IS0 460
I S0 460
5%
2.4%
Effluent water quality for the I S0 460 tests was
good, but degraded at the higher oil-water ratios.
This is due to the slow disengaging times
characteristic of viscous oils. However, oil
discharged with the water at this point in recovery
operations would be skimmed from the surface on
the next pass. The final water quality values given
for the last two samples should reflect the lower limit
of oil-in-water discharge. The oil effluent contained
less than 1.5% water on average which would have
little impact on the volume of storage required during
spill recovery.
Oil: 0.06% water
20 Water: 1200 ppmoil
Oil: 0.06% water
20 Water: 11 00 ppmoil
V-10 TESTS
Input Estimated Input
Oil Organic Flow
Phase (YO) Rate
comparable to the results obtained from V-5 testing
for recovered oil. The water quality, while still very
good, contained approximately twice the oil content
of samples fiom the V-5 separator. This is most
likely due to the added mixing in the annulus with
Effluent
Result
Table 3
V-10 Separator Test DATA
No. 2 Diesel - Water
I S0 460
I S0 460
I S0 460
(gpm)
- <5 3 Water: 550 ppmoil
- >10 3 Water: 2800ppmoil
5-10 3 Water: 510 ppmoil
Oil: 1.2% water
Oil: 1.5% water
I S0 460
I S0 460
I S0 460
I I Oil: 0.07%water
33% 20 I Water: 2200 ppmoil
(gpm)
- <5 3 Water: 550 ppmoil
- >10 3 Water: 2800ppmoil
5-10 3 Water: 510 ppmoil
Oil: 1.2% water
Oil: 1.5% water
I
I Oil: 0.14%water
9% 20 I Water: 1300ppmoil
Input Oil Input
Concentration Flowrate
Effluent
Result
I
I Oil: 0.05%water
30 I Water: 1200ppmoil 25%
2%
I Oil: 0.07%water
33% 35 I Water: 6200ppmoil
I
Oil: 0.3% water
Oil: 0.75% water
22 Water: 980ppmoil
I
I Oil: O.OlYowater
increased rotor diameter and shear forces on the
liquids. The V-10 rotor at 1300 rpm has a 1.5
times higher linear speed than that of the V-5 at 1700
rpm. The dispersion thus created is more difficult to
separate once inside the rotor. Data from separations
of San J oaquin crude oil while skimming are
presented in Table 4. As observed with the V-5
separation of a more viscous oil product from water,
higher contamination of the aqueous outflow results.
However, both oil and water effluent qualities are
still at reasonable levels.
Table 4
V-10 Separator Test DATA
San Joaquin Crude
I (gpm) I
20% I 25 I Water: 4500 ppmoil
I
I Oil: 0.05%water
17% I 21 I Water: 3500ppmoil
The next larger separator, the V-10, was also tested
during prototype development. The results of tests in
which No. 2 diesel were skimmed from an 8x30
test tank are given in Table 3. The rotor speed for
these studies was 1300 i 20 rpm which again
provides 200 gs of force. Separation quality is
24
V-20 TESTS
r
Input Oil Input Separator Emuent
Concentration Flow Analytical Result
Rate
Testing of the V-20 separator at flow rates above 100
gpm provided unique challenges. The test tank
volume of 2500 gallons is rapidly circulated (about
three times per hour of testing), and output flow
ratios are difficult to measure accurately while
sampling with present equipment. For these reasons,
no effort was made to measure the input oil
concentration while skimming at these flowrates.
Instead, 50 to 100 gallon spills of the test oil were
processed and sampled chronologically to
demonstrate typical effluent characteristics. The data
taken from tests of No. 2 diesel, ISA 460. gear oil,
and San Joaquin crude are given in Table 5. Again,
an increase in the oil content of effluent water is
observed as additional mixing in the annulus occurs.
The V-20 rotor speed tested was 850 rpm, but the
linear rate is 1.3 times that of the V-10 separator.
Again, this rotor speed provides 200 gs of force for
separation. Still, very favorable separation results
were obtained even at flow rates of 125 gpmand a
wide viscosity range of test oils.
(No. 2 Diesel)
As Skimmed-Test #2
(No. 2 Diesel)
As Skimmed-Test #3
(No. 2 Diesel)
As Skimmed-Test #4
(IS0 460 Gear Oil)
As Skimmed-Test #5
(IS0 460 Gear Oil)
As Skimmed-Test #6
(IS0 460 Gear Oil)
As Skimmed-Test #7
(IS0 460 Gear Oil)
As Skimmed-Test #8
Table 5
V-20 Separator Test DATA
Various Oil - Water
oil: 0.1% water
Oil: 0.05% water
Oil: 0.05% water
Oil: 4.0% water
Oil: 2.3% water
Oil: 1.0% water
Oil: 3.4% water
125 Water: 9300 ppm oil
125 Water: 8600 ppm oil
125 Water: 5000 ppm oil
125 Water: 9600 ppm oil
125 Water: 12,000 ppm oil
125 Water: 4800 ppm oil
125 Water: 2200 ppm Oil
(San Joaquin Crude)
As Skimmed-Test #10
(San Joaauin Crude)
I (gpm) I
As Skimmed-Test # I I 125 I Water: 3800 ppm oil
Oil: 0.3% water-
Oil: 0.1% water
127 Water: 3900 ppm,oil
(IS0 460 Gear Oil) I I Oil: 3.0% water
As Skimmed-Test #9 I 127 I Water: 12,000 ppm oil
occurred over the past 18 months. The design is
elegant in simplicity employing only one moving
part. Results are favorable when employed as an
emergency cleanup device for primary oil-water
separation. Even the V-20 is compact and
lightweight enough for airlift delivery to remote
locations. It is easily deployed on modest sized boats
and simple to operate. Plans for continued testing of
this separator in actual oil spill recovery events are
being finalized. Modifications to the design aimed at
lowering the mixing of water and oil within the
annulus are in progress. Further scale up to a V-30
with an expected throughput of greater than 400 gpm
has been completed. Fabrication and testing of this
first prototype will occur in the near future.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the contributions
of co-workers Pete Harris, Todd Lafrenz and Dan
Cornel1 whose efforts in many phases of this
development are much appreciated. Our thanks also
to Mary Manning and Mary Horn for their diligence
in preparation of this manuscript. Finally, thanks to
Bob Geminder for his inspiration and
encouragement.
REFERENCES
Lindsey, A.W., Jakobson, K., An Overview
of the EPA Oil Spill Research Program.
Proceedings of the 1993 Oil Spill
Conference, American Petroleum Institute,
Washington, D.C., pp 545-547, (1993).
Tennyson, E.J., Results from Oil Spill
Response Research - An Update. Proceedings
of the 1993 Oil Spill Conference, American
PetroleumInstitute, Washington, D.C., pp 54 1-
544 (1 993).
Meikrantz, D.H., Method for Separating
Disparate Components in a Fluid Stream,
U.S. Patent Number 4,959,158,
September 25, 1990.
SUMMARY
Development of a new high volume centrifugal
separator with application in spill recovery has
25