Hydrocyclone A Solution To Produced Water
Hydrocyclone A Solution To Produced Water
Hydrocyclone A Solution To Produced Water
Produced-Water Treatment
N. Meldrum, * Conoco U.K. Ltd.
Summary. Progressing into deeper water and more hostile environments in the search for new oil and gas reserves has placed
an increasing demand on the industry to develop lighter, more compact, and more efficient process equipment to replace their
traditional counterparts. A recent application of cyclone technology for liquid/liquid separation of oil from produced water has
shown considerable promise during extensive field testing. This paper outlines the basic construction and principle of operation of
the deoiling hydrocyclone and discusses system design, early operational expe~iences, and test resu~t~ from the first full-scale
commercial application of the four-in-one hydrocyclone concept on the MurchIson platform. In addItIOn, and of perhaps more
significance, early results from field tests of the larger 60-mm [2A-in.] cyclones on the Hutton tension leg platform (TLP), where
conventional equipment was adversely affected by platform motion, are discussed.
Introduction
Produced water represents an increasing portion of the total pro- This is largely an equipment-based standard, based on current tech-
duced fluids during the life of a field. Although the treatment of nology and using well-operated and well-maintained separation
produced water offers no direct economic incentive, a considerable equipment. A further concession of an upper limit of not more than
portion of both engineering and operations time is spent modifying 4 % of the samples exceeding 100 mg/kg recognizes that control
and maintaining produced-water cleanup systems to ensure satis- equipment is not infallible and upsets do occur.
factory operation.
In the Murchison field, although the existing system performed Construction and Principle of Operation 1·3
adequately, albeit through labor-intensive operation, optimum Construction. The functional geometry of the Vortoil hydrocyclone
reservoir development required water-injection capacity beyond the (i.e., involute inlet/swirl chamber, concentric reducing, fine taper,
original design basis. Consequently, this created the need to up- and parallel tail sections) is fabricated as a liner and contained within
grade the capacity of the produced-water treatment system. A similar a pressure-retaining outer shell (see Fig. 1). The outer shell is de-
upgrade in water-handling capacity was required for the Hutton signed in accordance with internationally recognized codes and is
field, but more important, the platform could not achieve the U.K. constructed from essentially standard piping components. It can
Dept. of Energy specification for overboard effluent of 40 ppm with therefore be designed easily for a variety of pressure ratings and
the originally installed equipment. Investigations into alternatives could eventually become an off-the-shelf item. While there are ob-
to the traditional flotation cells/plate separators for oily-water vious advantages with standardized designs, almost unlimited flex-
cleanup revealed that the hydrocyclone separator warranted serious ibility is available for nozzle positioning and orientation, which
consideration. should simplify the piping engineer's task, particularly during a
Our first experience with hydrocyclones was with pilot plant trials retrofit where considerably more constraints will be imposed than
of the Serck-Baker Oilspin developed as a result of an extensive in a new design situation.
research program conducted at Southampton U.1-3 While these A considerable degree of flexibility exists with the hydrocyclone
tests, carried out on the Murchison and Hutton platforms in early liners, which can be fabricated in either single or four-in-one 35-
1985, were not an unqualified success, they demonstrated' the and 60-mm [104- and 2 A-in.] units. The size refers to the diameter
equipment's potential and enough encouragement for us to commit at the transition between the concentric-reducing and fine-taper
to the first full-scale commercial application of the Vortoil® four- sections of the cyclone. This provides an almost unlimited turndown
in-one 35-mm [lA-in.] hydrocyclone later that year. capability that can handle the extremely low water volumes expected
the Vortoil hydrocyclone, developed by an Australian company during early operations or as part of a test separator water-treatment
but based on the original Southampton U. design, was considered package where significant flow variations are always apparent. As
to be a more refined and advanced product capable not only of water cut increases or, as in the case of Murchison and Hutton,
achieving the required performance but also of reducing the nurse- the originally installed water-handling capacity is upgraded, addi-
maid attention associated with more traditional equipment. The many tional units can simply be added as a bolt-on expansion.
operational advantages of Vortoil make it ideally suitable for a The liners are fabricated in stainless steel with a cast stellite inlet
retrofit situation. However, the true worth of hydrocyclone tech- for improved erosion resistance. All liner components are readily
nology should be realized on future marginal developments where replaceable on an individual basis, and with no moving parts, main-
space and weight are a premium and on floating structures like the tenance should be relatively simple with long periods between sub-
Hutton TLP where motion sensitivity is a problem. sequent inspections once the erosion resistance for a particular
application has been confirmed.
U.K. Legislation
The disposal of produced water into the British sector of the North Principle of Operation. Separation is achieved by generating cen-
Sea is governed by "The Prevention of Oil Pollution Act, trifugal forces that are orders of magnitude higher than available
1971. "4,5 In summary, this act makes discharging any oil into the in conventional gravity-based separation equipment. A high-velocity
sea an offense; however, the secretary of state has the power to vortex with a reverse flowing central core is set up by entry through
grant exemptions from the act, which provides the necessary oper- a specially designed inlet into a cylindrical swirl chamber. The fluid
ational flexibility. is accelerated through the concentric-reducing and fme-taper sections
The Dept. of Energy administers the legislation and has the power of the cyclone, where the bulk of the separation occurs, into the
to enforce the agreed target standard of 40-mg/kg monthly average. parallel-tail section, where the smaller, slower-moving droplets are
recovered.
'Now at Dubai Petroleum Co. The lighter oil droplets migrate toward the lower-pressure central
Copyright 1988 Offshore Technology Conference core where an axial reversal of flow occurs, resulting in the lower-
SPE Production Engineering, November 1988 669
100 .. . ."
...
. .. .
I
0
..
o.
;!' •
"• •
0
o ,'" o •
90
..· .
0
'
. ·· ....--.
80
:
1
70 ·
--
:I!.
~ 60
>
0
Z
W 50
(3
it
LL. 40
W
30 \
• 60mm UNIT cln > 1000 PPM
20 ... 35mm UNIT cln < 100 PPM
Fig. 1-Four-ln-one, 3S-mm hydrocyclone photograph (cour- • 35mm UNIT cln > 1000 PPM
tesy BWN Vortoil Pty. Ltd.).
10
.. .. o~~~~~--~~~~l
..'.. .
100
++!+..
'.r • . .
o ~ ~ ~ ~1~1~1~1~
+ +
.. .•
+ + +
90
... . .
•
FLOWRATE BWPO/4 IN 1 UNIT
..
Fig. 3-Four-ln-one hydrocyclone efficiency vs. flow rate .
80
+
...• • .
.f:
70 CONCENTRATIONS Hence, it can be seen that these are the major factors affecting
•
--
:I!.
~
>
60
+ ~ E = C 1n • COUI
c 1n
separation .
1. The specific-gravity difference, ~'Y, is the driving force for
separation. The greater the difference, the greater the potentilil for
• c out
0
Z
50 +
• FLOWS
rapid phase separation. _
W 2. The mean droplet size, d, is important because larger droplets
(3
it
LL. 40
w
qr.l~ I' qw
REJECT ~
RATIO qw
move more rapidly toward the central core. Higher inlet oil con-
centrations increase the probability of coalescence giving rise to
an increased mean droplet size. Additionally, careful design to
minimize high-shear areas upstream of the hydrocyclone is important
30
to prevent droplet disintegration .
Ie • 35mm, c 1n < 100 PPM
... 35mm, c 1n > 1000 PPM 3. Temperature, T, affects both density and viscosity, increasing
20 + 6Omm, c 1n > 1000 PPM interphase migration velocity. Installation downstream of the
primary separator where 95 % water removal typically occurs but
10 upstream of any interstage or produced-water coolers provides the
r. best performance.
4. Higher flow rates, q, increase the intensity of the centrifugal
0 separation forces.
0 1 2 3 4 5 6 7 8
As mentioned previously, the level of reject rate is an important
REJECT RATIO (%)
parameter affecting performance. However, providing that it is
maintained above the characteristic minimum (discussed later), it
Fig. 2-Hydrocyclone efficiency vs. reject ratio.
doesn't significantly influence performance and can therefore be
treated as a constant.
The reject rate and inlet flow rate provide the means of efficiently
density oil-enriched phase being removed through a small-diameter controlling the hydrocyclone for optimum performance over a range
orifice in the center of the inlet head. The oil-depleted water stream of conditions. While the other parameters are undoubtedly equally
exists as the continuous phase from the downstream end. It is also important, they are essentially fixed for a given installation, so we
worth noting that there is no inward-projecting vortex finder, unlike do not have the same degree of control over them, except through
the conventional hydrocycJone, because re-entrainment carry-over careful system design.
merely causes dilution of the reject stream.
Reject Ratio. The reject ratio, expressed as a percentage, is de-
Factors That Influence Performance. The oil-removal efficiency fined as the ratio of fluid rejected as concentrated oily water from
of the deoning liquid/liquid hydrocycJone is influenced by a number the upstream outlet to the clean-water underflow. As the reject ratio
of factors, some of which currently are not fully defined or under- increases, the efficiency of separation increases and then plateaus
stood. It is certain, however, that the inlet flow rate, reject rate, and remains essentially constant. This is presented graphically in
droplet size, oil ~oncentration, differential specific gravity, and tem- Fig. 2, which shows the characteristic curves of efficiency vs. reject
perature play significant roles. This can be represented simplisti- ratio for Murchison and Hutton with a well-defined breakpoint at
cally by the following approximation: approximately 1 %.
In some cases, further increases in reject rate may yield a mar-
E=f[(4 'Yw =NRe)d, ~'YJ . ...................... (1)
Q ginal improvement in separation efficiency, but this is not normally
necessary or justified because of the increased volume of clean water
7rdjl d 'Yw
that would be recycled for subsequent retreating.
670 SPE Production Engineering, November 1988
14
26
13
12
24
b 11
~ 10
22 ~
• Pout ...
Q.
9
8
20 o
II: 7
., t:.P2
C
W 6
~ 18 c=---
t:.p,
II:
;:) 5
• IJ)
IJ) • 6. p. 3% REJECT
~ w
~
c.. 16 If 3
• 6. p. 2% REJECT
• 6. p. SHUT IN
0 • . 6.p1
a: 14
C
W
t:.p, =Pin - PreJ O+---'-~~~~---.----r---~--'----r--~--
a: o 2000 4000 6000 8000 10000 12000 14000 16000 18000
:::I 12
en
en
w 10
a: Fig. 5-Four-ln-one, 60-mm hydrocyclone pressure drop vs.
c.. flow rate.
8
6
6.0
4 5.5
5.0
2
NOTE: t:. P, - 1% REJECT RATIO 4.5
-s. 4.0
O+---~~~--~--~--r---r---~~~ 0
o 1000 2000 3000 4000 5000 6000 7000 8000 i=
e(
3.S C = PRESSURE DROP CONSTANT
II:
3.0 Pin-Pre!
to-
0 Pin -Pout
Fig. 4-Four-in-one, 35-mm hydrocyclone pressure drop vs. ..,
W 2.S
flow rate. W
II: Pin
2.0
1.S +
In most cases, Murchison and Hutton being no exception, very PreJ~
1.0
high oil-removal efficiencies can be achieved with a reject ratio
0.5
of - 1 %. However, while this holds true for the majority of cases,
the required reject rate is obviously influenced by the level of inlet 0
1.0
0 2.0 3.0 4.0 5.0 6.0
contamination and may be related to the volumetric throughput and ·c·
centrifugal separation forces. At lower flow rates, lower specific-
gravity differences, or with small droplet sizes, it is likely that the Fig. 6-Pressure-drop constant vs. reject rate.
central core will be less dense and some of the smaller, slower-
moving droplets would not be captured. In these cases, an increase
in the reject rate beyond the characteristic breakpoint previously the reject stream. This inhibits the reject flow rate and results in
mentioned is likely to be of more benefit. little and eventually no separation.
The low level of reject rate is a significant advantage hydrocy- On both Murchison and Hutton TLP, the reduction in separation
clones have over conventional equipment because this volume of efficiency at high operating flow rates has been the result of a re-
fluid requires further treatment before disposal. However, perhaps stricted reject flow rate, not droplet shearing. This does not mean
of more importance is the ease of retreatment of the reject stream, that the effects of droplet shearing are insignificant, but at the flow
which splits rapidly into free oil and water. By comparison, the rates we operate at, little or no droplet shearing occurs'in the cy-
launder stream from the conventional flotation cell, which is highly clone itself. It is therefore important to establish the operating range
chemically contaminated, forms a stable emulsion that on the Hutton of flow rates over which the required performance can be consis-
TLP has proved particularly difficult to treat. tently obtained. The ratio of maximum to minimum operating flow
rates that yields the necessary performance is called the turndown
Flow Rate. The inlet flow rate is another key operating character- ratio.
istic of the hydrocyclone. Typically, as flow rate increases, the ef-
ficiency of separation increases and then levels out over the unit's Control. The control scheme for a hydrocyclone installation in
operating range. Further increases in flow will eventually cause which we have identified flow rate as a key parameter would logi-
the efficiency to drop sharply (Fig. 3). cally be based on total flow rate with a ratio device adjusting a flow-
A certain minimum flow rate is necessary to set up the vortex control valve in the reject line to maintain the desired reject ratio.
motion and to establish centrifugal separation forces that grow in However, problems can be expected with an installation based on
intensity as the flow rate increases and improve the separation ef- flow/ratio control because of gas breakout, which exists preferen-
ficiency. A point will be reached for a given set of conditions, tially with the reject stream and creates two-phase flow. Normal
however, beyond which further increases in flow rate will cause metering devices will not tolerate two-phase flow and give suffi-
performance deterioration. This is caused principally by the fol- ciently accurate and reproducible results for control purposes.
lowing. Tests with mass flowmeters, which are tolerant of two-phase flow
1. At very high flow rates, shearing of the oil droplets can occur, up to about 30% mixture, have been moderately successful and could
creating smaller droplets that are harder to separate. be used in some instances with careful design and sufficient system
2. As flow rate increases, the core pressure approaches at- backpressure to restrict gas breakout. Because the maximum
mospheric pressure, thereby reducing the available pressure to drive throughput is obtained with atmospheric pressure in the reject
r--T'!.IIL PT
C-1101 LEVEL CONTROl~ _ ~ ___ _
I I FROM C-t101 ~ T""
I I
I I flOW CONTRo'6 ~. :
Q _0
I I
: ~-----
I RATIO
_ _ _ -""'- _ _ _ _
I--.,...---o{:x:x:}---_r"--I
_ _ PT I""'i-:-{>I<:]-.., CLEAN
L...,
: Be(. . ."'"
WATER
Pln- P_
OVERBOARD
L ~SETPOINT
-~--1
35-mm Hydrocyclone
Reject Inlet-Oil Outlet-Oil
~L ~ ~ ~ .~.~.
Flow Rate Ratio Concentration Concentration Efficiency
(BWPD) (%) (ppm) (ppm) (%) >60
!
(J
1,812 1.2 48 26 45 iljso
1,963 2.4 48 13 73 U ;; .....
2,114 4.7 62 13 79 ...il:4O
·l.~
2,416 1.5 52 13 75 W
30 • 60mm HYOROCVClONE
4,031 1.2 48 9 81 • 60mm HYDROCVClONE PUMPED
20 • 35mm HYDROCYClONE HIGH INLET
4,243 1.0 52 10 80 CONCENTRATION
II 3Smm HYDROCYCLONE LOW INLET
4,530 4.3 35 5 86 10 CONCENTRATION 01020304050801080.'00
5,300 1.0 34 9 74
O+----r----r----r----r---~--~----~----
5,995 0.8 64 20 69 o 1000 2000 3000 4000 5000 6000 7000
6,010 1.0 25 7 72
6,010 3.1 29 3 89
6,070 1.2 30 5 83
6,433 1.1 34 8 76 Fig. 8--Hydrocyclone efficiency vs. Inlet 011 In water concen-
6,908 1.2 58 9 84 tration.
7,424 0.6 31 13 58
7,826 <0.1 54 48 11
35-mm Hydrocyclone High Inlet Performance Evaluation-Murchison
2,914 4.0 2,304 9 99 The Murchison system was commissioned during late Dec. 1985
2,944 12.0 15,700 6 99 and achieved high levels of efficiency and within-specification water
2,975 4.0 408 17 96
9 99
immediately upon startup, confirming the system's simplicity of
2,990 17.0 82,600
4,047 -- 132,000 573 99 operation and control.
4,077 -- 99,000 1,304 99 Because this was the first worldwide application of the four-in-
4,077 -- 139,800 8,000 94 one 35-mm [l.4-in.] hydrocyclone concept, a rigorous test program
4,107 -- 110,000 15,000 86 was developed to confirm its suitability for the demanding North
5,964 1.8 5,581 27 99 Sea environment and to advance the knowledge gained from earlier
5,995 2.0 13,773 19 99 pilot -plant trials in Bass Strait, Australia.
6,010 1.9 12,032 21 99 This was split into three stages: initial shakedown, establishment
6,010 2.1 2,765 17 99 and demonstration of the fundamental characteristics and operating
6,896 0.7 5.376 69 99
25 99
parameters, and examination of the impact of the platform oper-
6,921 0.9 3,328
ating variables on performance and the impact of the hydrocyclone
60-mm Hydrocyclone on other platform systems. The initial shakedown essentially con-
3,568 2.4 105 9 91 sisted of bringing the system on line, establishing a reliable oper-
6,593 2.0 313 15 95 ating base, and confirming that all components were functioning
7.336 3.1 370 20 95
properly. The emphasis during this time was on confirming that
7.336 4.1 846 18 98
4.6 23 98
the overboard specification could be achieved without having nec-
8,695 1.007
9,600 3.8 680 31 95 essarily optimized performance.
10.312 3.1 3.340 48 99 The next step was to determine efficiency as a function of reject
10.778 3.1 4.144 45 99 ratio while holding the inlet flow rate approximately constant. A
10,868 3.0 5.780 69 99 series of runs !tas been completed with reject ratios in the range
10,868 3.0 5.738 212 96 of 0.1 to 17.0%.
10.959 3.2 2.040 43 98 From inspection of typical results presented in Table 1, it can
11,140 2.5 2,153 35 98 be seen that high efficiencies can be obtained with a reject ratio
12,136 2.1 1,870 27 99 of - 1 %. This is presented graphically in Fig. 2, which shows the
12.680 2.1 2,168 34 98
characteristic curve of efficiency vs. reject ratio with a well-defmed
12,952 1.6 5,980 212 96
15.215 0.7 1.084 78 93 breakpoint at - 1 %. For normal operating conditions, it has been
confirmed that further increments in reject flow rate serve only to
60-mm Hydrocyclone Pumping Test
dilute the reject stream and to increase the amount of fluid recycle
5.832 7.2 1.510 188 88 for subsequent retreatment.
7.245 5.1 2,009 212 89 The efficiency of oil removal can be seen to rise sharply as the
9,057 2.0 2,876 190 93
reject ratio is increased from 0 to 1 %. However, further increases
9.057 2.0 3.088 256 92
9.057 1.9 4.585 230 95
provide no significant performance improvement. The plateau ef-
9,600 3.1 1,988 239 88 ficiency in this series of runs was - 85 %. This would normally
9,782 3.4 4.125 226 94 be considered a low efficiency for the hydrocyclone, but the average
9,872 2.4 7,498 348 95 inlet concentration for the above runs was 40 ppm, with outlet con-
10.959 2.1 920 162 82 centrations typically ranging from 5 to 10 ppm. Inspection of the
10.959 2.0 1.690 207 88 inlet samples showed slight discoloration of the water but no visible
11,774 3.0 2,475 195 92 free oil. The corresponding outlet samples were clear to the naked
11,774 3.4 2,644 230 91 eye and could easily have been mistaken for drinking water. Further,
11.865 0.9 4,600 315 93
the dissolved oil content on Murchison accounts for about 5 ppm
12.861 1.3 1.218 315 74
13.586 2.2 2.398 250 90 alone. Hence, the hydrocyclone was performing exceptionally well,
14,582 1.3 3,215 305 90 cleaning up clean water that probably consisted of a large amount
of very small droplets that are not easy to remove.
It was considered necessary, however, to expose the cyclone to
wider and various conditions to confirm that it could handle higher
levels of inlet contamination. Its response to process upsets, when
a valve failure could result in a slug of oil being carried out in the
produced water stream was also tested.