Separation and Purification Technology 134 (2014) 241–246

Contents lists available at ScienceDirect

Separation and Purification Technology

journal homepage: www.elsevier.com/locate/seppur

The size and performance of offshore produced water oil-removal

technologies for reinjection
S. Judd a,⇑, H. Qiblawey a, M. Al-Marri a, C. Clarkin b, S. Watson b, A. Ahmed c, S. Bach c
a
Department of Chemical Engineering, Qatar University, Doha, Qatar
b
Maersk Oil Qatar, Doha, Qatar
c
Maersk Oil Research and Technical Centre, Doha, Qatar

a r t i c l e i n f o a b s t r a c t

Article history: Produced water (PW) is wastewater generated from oil exploration, and requires treating for oil and sus-
Received 21 June 2014 pended solids removal. The viability of an effluent treatment unit process for this duty is dependent both
Received in revised form 11 July 2014 on its efficacy, in terms of oil removal and – for offshore applications especially – its size, in terms of its
Accepted 12 July 2014
area (FA, m/h) and volume (FV, h1) footprint per unit volume flow. The incurred footprint applies to both
Available online 29 July 2014
the individual unit (vessel, column or tank) and the collection (or array) of units/vessels in a skid.
An assessment of unit process footprint based on available information has been conducted, in partic-
Keywords:
ular to the case where high-quality treated water is required for reinjection. The analysis encompasses
Produced water
Oil
technical data from specific proprietary technologies as well as generic information for process technol-
Footprint ogy types. Technologies considered comprised hydrocyclones (HCs), induced gas flotation (IGF), media
Hydrocyclone (nutshell) filtration (NSF), and crossflow membrane filtration (CMF).
Flotation The analysis revealed the HC to incur the smallest area footprint, less than half that of an IGF, notwith-
Filtration standing only 0.15% of the total skid volume being used for the actual separation process. The CMF had a
slightly smaller area footprint and less than half the volumetric footprint of the NSF, if the requirement
for backflushing is considered. The fitting of the modular HC and CMF technologies to a skid incurs a con-
siderable increase in the footprint, particularly for the HC where the volume occupancy is increased by an
order of magnitude. It was concluded that spatial efficiency gains could be attained for modular processes
if spacing of the HC vessels or membrane modules can be reduced, contributing significantly to the via-
bility of CMF in particular.
Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction
Produced water (PW) generated by the oil and gas industry rep-
resents the largest volume by-product of petroleum production,
with a water to oil ratio which increases with increasing well
age. Current estimates of the average ratio of water:oil globally is
4:1 [29], with generation of produced water offshore likely to
increase more rapidly than that onshore. The latter arises because
offshore wells are operated for longer to offset the capital invest-
ment, such that the water content is generally higher. Whilst PW
contains a number of different chemical species which may be
adverse to either environmental impact or oil platform/well
operations, the determinant of most importance is the residual
oil concentration which is largely present in the suspended form.
The threshold suspended oil concentration is limited either by
⇑ Corresponding author.
E-mail address: simon.judd@qu.edu.qa (S. Judd).
the legislated limit for discharge or, if it is to be considered for rein-
jection (PWRI), the permeability of the oil reservoir. For PWRI, the
concentration of other suspended materials (i.e. the suspended sol-
ids, SS) is also of significance.
The technologies selected for PW treatment depend on whether
the installation is based onshore or offshore (Fig. 1). For onshore
installations footprint is generally a less critical factor than off-
shore. Simpler, low-energy/high-footprint technologies may there-
fore be employed, and these may target both the suspended and
dissolved oil depending on the target treated water quality. Such
treatment processes can potentially include biological-based tech-
nologies adapted from those routinely applied to municipal waste-
water [16,23], more usually employed for ‘‘downstream’’ refining
processes [12]. The large footprint incurred by biological treatment
technologies, however, make them untenable on an oil platform
even when intensified as a membrane bioreactor (MBR). Aerobic
biotreatment also introduces dissolved oxygen, which can be prob-
lematic due to the corrosion incurred. Both aerobic treatment and

http://dx.doi.org/10.1016/j.seppur.2014.07.037
1383-5866/Ó 2014 Elsevier B.V. All rights reserved.
242 S. Judd et al. / Separation and Purification Technology 134 (2014) 241–246
Fig. 1. Offshore produced water treatment.
advanced oxidation processes (AOPs) are therefore only viable for
discharge purposes.
Environmental pressures have placed increasing emphasis on
PWRI, which is the only viable reuse option offshore and whereby
discharge to sea can be avoided. For fractured oil reservoirs, where
water can flow more freely without clogging the reservoir pores,
the water quality requirements are not onerous and the classical
two-stage process of hydrocyclones (HCs) and induced gas flota-
tion (IGF) is normally sufficient (Fig. 1). The IGF may be substituted
with a simple degassing vessel in some regions (such as the North
Sea), and degassers and/or surge tanks may also be used upstream
of the HCs. For ‘‘tight’’ or low permeability reservoirs – normally
associated with carbonate strata – removal of particles down to
3–5 lm in size may be necessary, demanding the use of filtration.
It is of interest to appraise the different suspended matter/
water separation technologies for PW treatment with specific ref-
erence to their incurred footprint with specific reference to off-
shore duties. These include both the classical primary and
secondary technologies along with the tertiary processes of media
and membrane filtration. Media filtration, if used, is normally
based on nutshell filters (NSFs). The more advanced emerging sep-
aration process of crossflow membrane filtration (CMF) provides a
more consistent quality effluent, but is generally considered to be
comparatively large in footprint.
2. Technologies
2.1. Fundamental parameters
Produced water is normally taken as being the water which
exits the production separator (Fig. 1), whose primary purpose is
to partition the oil and gas phases. Subsequent processes treating
the water phase may recover oil within a concentrate stream, but
their primary purpose may be considered as being wastewater
purification.
In terms of space occupancy the specific flow capacity is given by:
Capacity by area (flow/area) FA = Qave/A
Capacity by volume (flow/volume) FV = Qave/V
where Qave is the mean flow rate, A is the projected floor area of the
technology and V is the technology volume (either the individual
unit/vessel or the skid). FA and FV thus respectively take units of
velocity and inverse time. FA equates to the approach velocity for
column process such as media filtration or the horizontal flow
velocity for separator tank, such as an API vessel. FV is the inverse
hydraulic residence time. It is most convenient for FA and FV, the
difference between which is simply the height of the technology,
to take units of m/h and h1 respectively.
Classical primary and secondary technologies normally com-
prise a combination of gravitation, HC and flotation technologies
(Fig. 1); for such technologies the mechanism for oil removal is
based at least in part on the droplet buoyancy. The fundamental
relationships for the design of these technologies are thus derived
from Stokes Law [27], and contain the function d2Dq/l where d is
the oil droplet size, Dq the density difference between the oil drop-
let and the water, and l the water viscosity. It therefore stands to
reason that oil removal by such technologies is highly dependent
on d, as well as to a lesser extent on the oil fraction (heavy oil being
the most dense and thus the least buoyant) and temperature,
which affects l. These factors may then be expected to also influ-
ence the values of FA and FV. Pretreatment to chemically destabilise
the oil droplets (using flocculants) increases their propensity to
coalesce, and thus their removal by virtue of their increased buoy-
ancy. Against this, some production chemicals, corrosion inhibitors
in particular, tend to stabilise oil droplets making their removal
more difficult. Other key factors affecting process efficacy include
the presence of suspended solids, which actually determines the
operating cycle for some unit processes (Section 3.1).
2.2. Data capture
Very little useful data is available from the peer-reviewed liter-
ature. Data were captured largely from grey literature sources
(supplier information, internet and papers from specialist confer-
ences) as well as via personal contacts (technology suppliers, con-
sultants, contractors and end users) and text books (Table 1). The
technology volume in the analysis was taken to be based on either
its single principle component (column, tank, array, etc), or a skid/
array comprising normally a number of such units or vessels. The
inclusion or exclusion of ancillary equipment (pumps, control pan-
els, etc) was also considered, along with the requirement for duty
and standby components (as would be the case for significant peri-
ods of downtime). Finally, the sensitivity of sizing, and selection of
the unit operation generally, to technology-specific factors was
identified.
FA data for API (American Petroleum Institute) separator and
NSF units were obtained directly from design guidelines, reference
texts [14,27] or technical papers. For these technologies the maxi-
mum horizontal velocity Vh (FA = height  Vh/length) in the case of
the API separator [2] and approach velocity Va (or hydraulic loading
rate, HLR), for the NSF [3] respectively indirectly and directly
equate to FA. Footprint data for all other technologies were calcu-
lated from their dimensions and flow capacities, as provided by
the various information source types indicated in Table 1. This
includes data for skid-mounted technologies, as provided from
product brochures and/or end users. Performance data (oil
 S. Judd et al. / Separation and Purification Technology 134 (2014) 241–246 243

Table 1
PW unit operations considered and data sources.

Technology Specific type No. data Ref. booka Techn. suppl. End user Report/paper
Separators API separator 2 1 1
Coalescers Corrugated plate interceptor (CPI) 2 3
Hydrocyclones (HC) De-oiling hydrocyclone 5 2 3
Flotation, induced gas (IGF) Hydraulically-induced 12 9 3
Mechanically-induced 10 10
Compact flotation unit (CFU) 1 1
Filter, media Nut shell filter (NSF) 2 1 1 1
Filter, membrane Ceramic cross-flow (CMFb) 3
a
[27].
b
Cross-flow microfiltration or cross-flow membrane filter.

removal) was captured primarily from various conference presen-
tations or company reports.
The ratio of the space occupancy of the NSF columns on the skid
compared to the single column unit was assumed to be the same as
that calculated for an IGF skid and unit based on site data (1:2.3 for
FA; 1:6.9 for FV. It was further assumed in the analysis that the
water recovery from all processes exceeds 95% in all cases, such
that no correction was made for recovery.
3. Results
3.1. Performance data (suspended oil removal)
For the gravity separator, CPI, de-oiling HC and IGF data there
is an expected relationship between performance and droplet
size; data provided refer to removal without chemical pre-
treatment to enhance coalescence. There is also an impact of feed
concentration, with % removal increasing with feed concentration
– as arises with most water purification processes. For some tech-
nologies there is further variation due to other specific facets.
Removal by an IGF is dependent on the total number of cells, with
3–4 being the usual number employed. In the case of the HC the
minimum size of droplet removed increases with operating pres-
sure, due to the increased centrifugal force applied. For example,
according to the data provided for the Enhydra HC [5] the small-
est droplet removed decreases from 15 lm at 3 bar, to 8 lm at
20 bar for >95% removal. The deoiling HC technology is also the
only one of those listed which provides minimal suspended
solids (SS) removal, for which a dedicated de-sanding HC must
be used.
The two filtration-based technologies (NSF and CMF) both offer
extremely effective removal (>98%) of both the suspended solids
and free oil at very low particle and droplet size. However, both
are constrained by the feedwater suspended particle and oil con-
centration. The NSF is limited by the bed capacity for both the oil
(0.5–0.8 g per g media according to [26] and, more typically, the
SS. For a maximum solids loading of 54 kg/m2 and HLR of 64 m/h
[3] a PW containing 200 mg/L would require regeneration (by
backwashing) every four hours. Halving the HLR to that normally
recommended [20] doubles the run time but also the footprint.
Backwashing is a water-intensive process and incurs significant
downtime, such that standby capacity is needed which then fur-
ther increases the process footprint. The membrane process also
requires backflushing and is susceptible to fouling at high concen-
trations of free oil, which then reduces the capacity of the process
(via the permeate flux – the flow rate per unit membrane area)
consequently increasing the footprint.
There are additionally enhanced commercial coalescence/
interceptor technologies which are claimed to provide very signif-
icant removal. These include CTour by Prosep, which can appar-
ently remove suspended oil down to levels of 4–5 mg/L [28] as
well as some of the dissolved organics, and the TORR technology
[18] which removes particles as small as 2 lm (68% removal) to
10 lm (90% removal).

Table 2
Data summary, single unit, PW treatment.

Technology FA (m/h) FV (h1) Cin (mg/L) Cout % (rem) Min. drop dia. No. data Source(s), see Table 1
(mg/L) (lm) and note 1 below.
Onshore
API gravity separator 4.4 1.1 5000–20,000 50–100 80–90 150 1 [2]
Corrugated plate interceptor (CPI) 6.5–14 3.2 5000–20,000 40 90–98 20–40 2 [1,11]
Offshore
Hydrocyclone (HC) 100–440 450 – 20–80 90–95 12–20 4 [7,8,27,19,25]
Induced gas flotation (IGF) 2.4–16 4–15 200–500 25–50 90–95 10–25 21 [27,19]
Compact flotation unit (CFU) 45–90 – 200–500 15–25 90–95 10–25 2 [4,6,13]
Onshore/offshore
Nutshell filter (NSF) 27–37 13–24 20–50 2–5 99 2 2 [3,6,20]
Crossflow membrane filter (CMF) 312 136 20–50 0 99 <1 3 [15,17,24]

NB
1. Minimum droplet size and other supporting data taken from [10,22,27,9,21].
2. API separator design data is based on the widely quoted maximum horizontal flow velocity limit Vh = 3ft/min. FA = hVh/L where h0.4 w and L > 5 w.
3. The IGF data is based on four units operating is series. FA = Va, the approach velocity (or hydraulic loading rate).
4. The minimum footprint for the NSF unit is based on duty/duty standby and an HLR of 33 m/h; the maximum is based on duty/standby and an HLR of 24 m/h.
5. The media column ancillaries are assumed to add 15% to the footprint and 25% to the volume; a column height of 4 m was assumed.
6. The membrane skid ancillaries are assumed to add 15% to the footprint and 25% to the volume.
7. The HC area-specific flow data varies widely, from 100 to 440 m/h depending on flow capacity [8] for the unit and from 8 to 29 m/h for a single-unit skid [7].
8. Siemens now Evoqua. EPCON now part of M-I SWACO.
244 S. Judd et al. / Separation and Purification Technology 134 (2014) 241–246

Fig. 2. IGF data FA and FV data [27]: (a) hydraulically-induced and (b) mechanically induced, listed in order of decreasing flow capacity.

3.2. Footprint data
The footprint and performance data are summarised in Table 2
for published data. Data for IGF technologies specifically from a
single source [27] are shown in Fig. 2 (for both hydraulically and
mechanically induced technologies). Mean FA and FV values for
the four principal offshore technologies are depicted in Fig. 3a
and b for the individual units/vessels and the skids respectively
for data captured from the various sources.
4. Discussion
The IGF data set (Fig. 2) indicates the expected trend in increas-
ing specific flow capacity with increasing flow, as demonstrated for
both configurations alongside the HC data in Fig. 4. The trend exists
only for FA. The absence of any trend with FV is consistent with
performance being dependent on hydraulic residence time (HRT),
which is dependent only on the target performance in terms of %
removal of a given oil droplet size and not on flow capacity. The
decreased area footprint with capacity, on the other hand, reflects
the presence of ancillary components which incur a certain space
occupancy, such that the trends with flow have a positive intercept
in all cases.
HC technologies incur the lowest footprint. The area-specific
flow capacity FA of an HC unit is up to two orders of magnitude
greater than that of an API separator challenged with the same
duty, and it is additionally more selective for smaller oil droplets.
Whilst the process efficacy of HCs can be increased by applying
higher pressures, they are also perhaps the least tolerant of all
technologies to suspended solids, with moderate levels of SS
demanding pre-treatment post-treatment (often using another
dedicated HC).
There is a significant difference in the single unit/vessel and
skid data, as indicated in Table 2 and Fig. 4. The difference is great-
est for the modular processes, specifically the HC and CMF technol-
ogies, where fitting the modules within a skid increases the area
footprint by between 7 and 60 times depending on the data set
used. Unlike all other technologies considered, these are not based
on a single flooded vessel (such as a separator tank or a media
column) but on passage of the water through a multiplicity of
channels under conditions of controlled forced convection. These
channels – the HC liners or the CMF interstices – are then con-
tained within a cylindrical pressurised vessel of some fixed size.
Dimensions of an HC unit are normally within the range of 0.3–
1.5 m in diameter, and 1–2 m in length, whereas a ceramic mem-
brane module is usually much narrower (0.15–0.3 m in diameter).

Fig. 3. Footprint data, site-based and other data, (a) single unit/vessel, and (b) skid-mounted units/vessels for end user site-based data and all other data (largely from
suppliers).
 S. Judd et al. / Separation and Purification Technology 134 (2014) 241–246
Fig. 4. FA vs flow for IGF and HC technologies. IGF data refer to single units and
multiple suppliers; HC data refer to a single supplier for each trend.
An HC skid will normally comprise 4–10 units/vessels; the foot-
print incurred will thus depend on how closely these can be spaced,
which is often constrained by the pipework. In the case of the CMF
the skid might contain 20–100 modules held within a rectangular
frame, the modules being fed from a common header and the per-
meate collected in a common manifold. In this case the spacing is
constrained primarily by the requirement for the modules to be
accessible, since they may demand periodic replacement and/or
external cleaning. Fig. 3 suggests that the area footprint incurred
by a membrane skid is between that of the NSF and the IGF, based
on averaged data from three individual membrane technology sup-
pliers. However, both these modular processes at least in principle
offer the opportunity for intensification. Only 4% of the total volume
of the CMF skid actively conducts the separation process. In the case
of the HC skid the corresponding figure is 0.15%, based on an
assumed liner residence time of 1.5 s.
Of the remaining technologies, for which only data on the unit
process rather than the skid assembly has been captured in this
analysis (Table 2), the area footprint of the corrugated plate inter-
ceptor (CPI) is comparable to that of the IGF. The compact flotation
unit (CFU) has a significantly higher specific flow capacity. All three
of these processes would, along with HC technology, benefit from
chemical pre-treatment to enhance the droplet size; at droplet
sizes below 10 lm these technologies are all of limited efficacy.
In such cases further improvement in water quality, for example
for the purpose of reinjection into tight reservoirs, demands sup-
plementary treatment by a filtration technology, i.e. NSF or CMF.
These are roughly comparable in footprint to the routinely
employed IGF technology when skid mounted (Fig. 3). However,
the IGF process has an oil loading capacity considerably higher
(by an order of magnitude) than the filtration technologies.
5. Conclusions
An analysis of the relative footprints incurred by conventional
and novel candidate technologies for advanced PW treatment
indicates hydrocyclone (HC) technology to be the most intensive
in terms of the volume treated per unit floor area, as well as that
per unit volume. The individual HC vessels can process up to
440 m3/h of water per m2 footprint (i.e. FA = 440 m/h) or per m3
vessel volume (FV = 440 h1). This compares to figures of generally
between 7 and 10 for both FA and FV (in the same appropriate units)
245
for an induced gas flotation (IGF) technology. A crossflow filtration
membrane (CFM) filter module provides an FA of around 160 m/h.
The fitting of a modular technology to a skid incurs a consider-
able increase in the footprint. Only around 5% of the volume of an
HC vessel is dedicated to conducting the separation process, and in
a skid this figure decreases to 0.15%. For a membrane separation
technology the corresponding figures are around 80% and 4%.
Notwithstanding the loss of intensivity on skid-mounting, the HC
process yields the highest specific flow per unit area and per unit
volume of all the classical and novel technologies assessed. Given
that it also appears to achieve a similar performance to the IGF
in terms of the smallest droplet removed, it is postulated that
the most efficient PWRI design is one based on coalescence and
HC, with further polishing by CMF. However, a deoiling HC does
not remove suspended solids, such that a further unit operation
may be required as pretreatment (such as a desanding HC) incur-
ring additional footprint.
It is evident that spatial efficiency gains could be attained for
modular processes if spacing of the HC vessels or membrane
modules can be reduced. In the case of the CMF this must be
coupled with the maintenance of the appropriate permeate flux.
Acknowledgements
The advice and information provided by Maersk Oil Qatar, Alan
Hunton of HTS Consultants and Hank Rawlins of eProcess Technol-
ogies is acknowledged and gratefully appreciated by the authors.
The work was completed as part of project number QUEX-CENG-
CHE-13n14-03, Qatar University.
References
[1] ACS, Liquid–liquid coalesce design manual, ACS Industries LP, company
brochure, 2014.
[2] API, Design and operation of oil water separator. Publication 421, American
Petroleum Institute API, 1990.
[3] C.D. Blumenschein, Walnut shell filtration for oil and solids removal from steel
mill recycle systems, AISE Steel Technol. 78 (4) (2001) 33–37.
[4] J.T. Cline, Survey of gas flotation technologies for treatment of oil & grease, in:
presented at the 10th Produced Water Seminar, Houston, TX, Jan. 19–21, 2000.
[5] Enhydra Ltd, LL15 hydrocylone, company brochure, 2014.
[6] EPCON, CFU Technology, company brochure, 2006.
[7] eProcess, eProcess Technology engineering drawing, 2011.
[8] Exterran, P-CloneTM Hydrocyclone, company brochure PKLONE-08-04-EN-
DE12-US, 2012
[9] R. Gay-de-Montella, Water–HC separation in Alberta heavy oil extraction
processes, presented at the meeting of APEGA, Professionals in Engineering
and Geosciences in Alberta, Alberta, Nov 19, 2009.
[10] Hunton, Private communication, 2014.
[11] Hydrocarbon–Carbon, Filtration and Separation, application datasheet S1665,
2014.
[12] S. Ishak, M. Malakahmad, M.H. Isa, Refinery wastewater biological treatment:
a short review, J. Scientific Ind. Res. 71 (2012) 251–256.
[13] L. Jahnsen, E.A. Vik, Field Trials with EPCON Technology for Produced Water
Treatment, presented at the Produced Water Workshop, Aberdeen, Scotland,
March 26–27, 2003.
[14] K. Lee, J. Neff, Produced water: environmental risks and advances in mitigation
technologies, Springer Science and Business Media LLC, NY, 2011.
[15] Likuid Nanotek, Company brochure, 2014.
[16] A.R. Pendashteh, L.C. Abdullaha, A. Fakhru’l-Razi, S.S. Madaenic, Z.Z. Abidin,
D.R.A. Biak, Evaluation of membrane bioreactor for hypersaline oily
wastewater treatment, Process Safety Environ. Protect. 90 (2012) 45–55.
[17] P. Pedenaud, S. Heng, W. Evans, D. Bieonneau, Ceramic membrane and core
pilot results for produced water, OTC-22371-PP, paper presented at Offshore
Technology Conference, 4–6 Oct, Rio de Janeiro, 2011.
[18] M.J. Plebon, M.A. Saad, S. Fraser, De-Oiling of produced water from offshore oil
platforms using a recent commercialized technology which combines
adsorption, coalescence and gravity separation, in: 16th International
Offshore and Polar Engineering Conference, San Francisco, 28 May–2 Jun,
2006.
[19] C.H. Rawlins, Flotation of time oil droplets in petroleum production circuits,
presented at the SME and TMS meeting: Mineral process plant design – an
update, Tucson Sep 30–Oct 4, 2009.
[20] C.H. Rawlins, Performance and limitations of cyclonic and walnut shell
filtration technologies to oil and solids Removal from produced water. SPE
workshop Produced Water Handling, Terranea Resort, June 27–28, 2012.
246 S. Judd et al. / Separation and Purification Technology 134 (2014) 241–246
[21] J.T. Robinson, An overview of produced water treatment technologies, in:
presented at the 14th Annual International Petroleum Environmental
Conference, Houston, Nov 5, 2007.
[22] RPSEA, Ultra-deep discharge of produced water and/or solids at the seabed.
Report no, 019121–3100-01. Research Partnership to Secure Energy for
America, April 24, 2012.
[23] E.A. Sharghi, B. Bonakdarpour, P. Roustazade, M.A. Amoozegar, A.R. Rabbani,
The biological treatment of high salinity synthetic oilfield produced water in a
submerged membrane bioreactor using a halophilic bacterial consortium,
J. Chem. Technol. Biotechnol. 88 (2013) 2016–2026. 2013.
[24] Siemens SilverbackTM System. Product brochure, Siemens Water
Technologies, 2007.
[25] A. Sinker, Produced water treatment using hydrocyclones: theory and practical
application, in: 14th Annual International Petroleum Environmental
Conference, Houston, 2007.
[26] A. Srinivasan, T. Viraraghavan, Removal of oil by walnut shell media,
Bioresource Technol. 99 (2008) 8217–8220.
[27] M. Stewart, K. Arnold, Produced Water Treatment Field Manual, Gulf
Professional Publishing, NY, 2011.
[28] H. Torvik, L. Bergersen, C. Paulsen, One year of operational experience with
CTour at Statfjord C presented at the 3rd NEL Produced Water Workshop,
Aberdeen, Scotland, April 20–21, 2005.
[29] TUV-NEL, TUV 8th Produced Water Group Workshop, 23–24 June 2010,
Aberdeen, UK, 2010.