Desalination of Produced Water Using Reverse Osmosis

By Graciela Morales and Maria Barrufet , Texas A&M University

Many oil and gas wells, particularly those in mature fields, produce large amounts of
brine along with the hydrocarbons. Disposing of the brine can be costly, due to its
composition and large volume. For example, in the Permian Basin of West Texas and
New Mexico more than 490 million gallons of water per day are produced and re-
injected. The prospect of many millions of barrels of produced water from coalbed
methane wells planned for the Powder River Basin has complicated development of that
resource.

Historically, the oil and gas industry has not promoted on-site water desalination. The re-
injection or surface discharge alternatives were much less costly and there was little
demand for the water. However, growing demand for fresh water in many areas and the
development of lower-cost technologies for removing contaminants from water are
beginning to provide compelling arguments for produced brine desalination.

The Texas Water Resources Institute (TWRI) at Texas A&M University (TAMU)
currently is supporting a multidisciplinary program, led by the Department of Petroleum
Engineering to develop technologies to treat produced water and make it safe for use in
agriculture and wildlife habitat restoration. The aim of the TAMU project is the
development of small-scale, modular, transportable units capable of treating relatively
small amounts of brine inexpensively. The team will utilize new technology in solids and
oil removal and advances in remote process control to create units exhibiting low
maintenance and high reliability in the field. These small scale units will utilize
nanofiltration (NF) and reverse osmosis (RO) to remove contaminants from oilfield
brines.

Similar pressure-driven membrane filtration equipment installations are widely used in

desalination of brackish and seawater and compete successfully with traditional thermal
desalination operations. However, if RO is to assume a more prominent role in produced
water treatment, there is a need for sound engineering designs adaptable to modular
operations.

As one portion of this effort, the TAMU team has developed a static model using
parametric curves to allow scale-up of an integrated RO system. They are also developing
a dynamic model that will be the basis for a control system and automatic operation. This
article provides some basic background on RO systems in general, along with a brief
description of the static model.

RO Desalination Process
A typical RO water treatment system includes two primary process elements: a
pretreatment subsystem and the actual RO unit (Figure 1). Feed water quality determines
the amount and type of pretreatment necessary to make RO economical, and as such is
the limiting factor of most RO systems in operation today. Membrane surfaces are prone

1
to fouling by particulate matter, inorganic scales (i.e. carbonate and sulfates salts of
alkaline earth metals), oxides and hydroxides of aluminum and iron, organic material
(i.e., humic, tannic, etc.) and biological material (e.g. bacteria, fungi, algae).

A typical pretreatment unit consists of a sand filter, an activated carbon filter and a depth
cartridge filter. The sand filter is used to remove larger impurities, however, sand filters
can clog quite quickly and the relative coarseness of sand allows many smaller impurities
to pass through. (Osmonics Inc., Published in National Development 01, Mar 1992 –
Internet source).

The activated carbon filter absorbs low molecular weight organics and reduces the
amount of chlorine or other halogens, but does not remove any salts. This absorption
process takes time, so service rates are limited to a maximum of about 5 gpm/ft. The
accumulation of solids can require backwashing, however this can result in loss of the
relatively fragile activated carbon material. Over a period of months to years, the
adsorption capacity of the carbon diminishes, requiring replacement or reactivation, a
process not easily accomplished in the field. These filters may also need to be changed
periodically to avoid bacterial growth. Hydrocyclones, coalescing media, and organoclay
materials may also be used for the removal of oil in the pretreatment portion of these
systems.

In the depth cartridge filter, remaining particles (in the 1 to 100 micron range) are trapped
in the complex openings of a filter material constructed of cotton, cellulose, synthetic
yarns or “blown” microfiber such as polypropylene. These filters have a lower density on
the outside and progressively higher density toward the inside wall. The effect of this
graded density is to trap coarser particles toward the outside of the wall and the finer
particles toward the inner wall. These filters are often disposable. As particles
accumulate, the pressure drop across the filter increases and when the pressure difference
between filter inlet and outlet has increased by 5 – 10 psi relative to the starting point, the
filter is backwashed or replaced.

After pre-treatment to remove suspended particles, the incoming water is pressurized

with a pump to exceed the osmotic pressure (typically 200-400 psi, depending on the RO
system and the contaminants). A portion of the water (permeate) diffuses through the RO
membrane leaving dissolved salts and other contaminants behind with the remaining
water. This “reject” or “concentrate” is drawn off as waste. RO removes virtually all
organic compounds and up to 99 percent of inorganic ions.

RO membrane fouling is a complex phenomenon involving the deposition of materials on

the membrane surface rather than plugging of the system. Scaling of RO membrane
surfaces is caused by the precipitation of sparingly soluble salts from the concentrated
brine (especially CaCO3 and BaSO4). A number of chemicals may be added to prevent
membrane fouling. For example, sulfuric or hydrochloric acid is employed to reduce pH
and prevent CaCO3 precipitation. Sulfuric acid, while safer and less expensive than HCl,
will increase the content of sulfate ions in the feed water and consequently the risk of

2
CaSO4 precipitation. The addition of polyphosphates or, more recently, polycarboxylates
is employed for preventing CaSO4 scaling.

Chlorination, either continuous or at intervals, is a common pre-treatment method for

preventing the growth of bacteria and algae that may cause fouling in the system or
degradation of cellulose acetate membranes. The amount of chlorine required is
determined by the amount of organic matter in the feed water and by the water
temperature.

Reverse Osmosis
Osmosis, an integral part of the functioning of all living cells, is a phenomenon in which
a liquid (water in this case) passes through a semi-permeable membrane from a relatively
dilute solution toward a more concentrated solution. This flow produces a measurable
pressure, termed osmotic pressure. If, however, pressure is applied to the more
concentrated solution that exceeds the osmotic pressure, water flows through the
membrane from the more concentrated solution to the dilute solution (Figure 2). This
process, reverse osmosis, results in two streams of water: one relatively large volume
with a low concentration of dissolved impurities (permeate), and one relatively small
volume with a high concentration (reject).

Osmotic pressure, and thus the trans-membrane pressure required to overcome it, is a
function of contaminant component molecular weight and concentration. For example,
the osmotic pressure for a 2 percent by volume sodium chloride stream is 250 psi, while
the pressure for a 10 percent potassium chloride solution is 965 psi.

RO Membranes
During the last two decades significant advances have been made in the development and
application of microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse
osmosis (RO) processes. MF membranes reject suspended particles only, UF membranes
reject suspended particles and high molecular weight compounds, NF membranes also
reject low molecular weight compounds, and RO membranes also reject ions (Figure 3).

The membrane itself must be physically strong in order to stand up to high osmotic
pressure. Over 100 different materials are used to make RO membranes, however the two
most commonly used membranes are made from cellulose acetate (CA) and polyamide
thin film composite (TFC). These may come in spiral, tubular hollow fiber, plate and
frame, or proprietary configurations. Hollow fiber and flat sheet are the most commonly
used RO membrane configurations. Hollow fiber membrane is extruded like fishing line
with a hole in the center to create a tiny (100 to 200 micron) hollow fiber strand. Flat
sheet membrane, a continuous sheet rolled up like a large paper towel roll, is used in
spiral wound (SW) configurations (Figure 4). Although HF RO elements provide more
surface area, they are more prone to fouling (Table 1). The characteristics and
performance of these membranes differ as well (Table 2).
Amjad Z. and Zibrida J., 1998, Reverse Osmosis Technology: Fundamentals and Water
Applications. Association Water Technologies, Inc. Annual Convention, Washington,
DC.

3
Spiral membrane elements are loaded in a serial configuration in a pressure vessel (1 to 7
membranes per pressure vessel) and tubular membrane elements are loaded in a parallel
configuration in a pressure vessel (1 to hundreds of elements pre pressure vessel).
Multiple pressure vessels may be connected in a serial or parallel flow path.

If the product flux decreases to unacceptable values (typically >10% decrease) due to
membrane fouling, the membrane must be cleaned. The cleaning method and frequency
depend on the type of foulant and the membrane’s chemical resistance. Generally, it is
easier to clean a membrane that is slightly fouled. Cleaning methods include mechanical
cleaning (i.e. direct osmosis, flushing with high velocity water, ultrasonic, sponge ball or
brush cleaning, air sparging, etc.), chemical cleaning (use of chemical agents), or a
combination of both.

TAMU Sizing Model

The TAMU research team has developed a set of theoretically grounded parametric
curves (Figure 5) that enable a designer to quickly arrive at an optimal RO unit size and
operating scenario by following an iterative procedure. Primary inputs for the procedure
include: feed water ionic composition, salinity, temperature, and pressure. These are used
to calculate several intermediate parameters which are embedded in the equations plotted
in Figure 5. By assuming a delta pressure (∆P) and desired permeate flow rate (qP), one
obtains a value for membrane area (Am) from the left axis of the chart. Using this value,
and assuming a value for membrane length/area ratio (L/A), one determines a value for
feed flow rate (qF) from the right axis. This value may be too large exceeding the amount
of feed water available. This value is then checked against a tolerance range and if it is
not within tolerance, the procedure is repeated by changing either the proposed value of
L/A or ∆P, and iterating. The procedure ends when the feed flow (qF) is within expected
values, resulting in values for permeate flow rate, membrane area, and membrane
length/area ratio that correspond to a particular feed rate. Following this procedure one
can observe that, as might be expected, membrane area increases when increasing the
brine concentration for a given permeate flow rate. (Note: A comprehensive description
of the theoretical basis used to develop these parametric equations, as well as an example
of their application, is available from the author.)

Ongoing RO Research
In addition to the development of this set of parametric curves, the research effort
currently underway includes:
• Evaluation of different RO membrane configurations for various membrane types
• Investigation of the effects of operating conditions (pressure, temperature and
brine composition) on RO membrane flux
• Experimental determination of the best diffusivity model
• Investigation of the effects of temperature and composition on osmotic pressure
• Completion of a sensitivity analysis across a range for three key variables: salt
concentration and types of salts up to 200,000 ppm); temperature (100 to 200 °F)
and pressure (220 to 2,000 psi).

4
During the past two years the TAMU team has identified and tested a hybrid system
consisting of pretreatment methods, inexpensive centrifugation technologies and
ceramic/polymeric RO membranes. Preliminary results indicate that the brine treatment is
feasible and can be done simply and economically. The research team has been working
closely with PCI Membrane Systems and Somicon A.G., who have redesigned their
commercial membrane modules by changing membrane coating methods, polymer
combinations and membrane element configurations. In addition PCI and Somicon
developed new low fouling NF and RO membrane materials and low cost module
geometries.
Using specifically formulated membrane modules, actual process water tests were
completed at the university incorporating state-of-the-art pre-treatment technologies that
included organo-clays, inexpensive and selective centrifuges and microfiltration systems.
The test results show that the treated wastewater quality is equal or better than that of tap
water.
For additional information on the status of this work or to become involved in industry
efforts to support this research contact the author, Dr. Maria Barrufet, at 979-845-0314,
or via e-mail at barrufet@spindletop.tamu.edu. Or members of this research team
Dr. Sefa Koseoglu s-koseoglu@tamu.edu (membranes)
Dr. Graciela Morales gmorales@ciunsa.edu.ar (diffusión models)
Mr. David Burnet burnett@gpri.org (regulatory aspects)

5
 Figure 1: Schematic of a Typical RO Treatment System

Chemical
addition

Concentrate or
Brine Water Reject

Pump
5 µ m cartridge RO System
Activated carbon filter
Sand filter
filter Permeate

Figure 2: Schematic of Osmosis and Reverse Osmosis

OSMOSIS REVERSE OSMOSIS

Head = Osmotic
Pressure
Applied
Pressure
Semi-permeable
Membrane
Diluted Brine
Initial Brine or pure water Concentrated Brine

Water Flow Water Flow

6
Figure 3: Pressure Driven Separation Processes

Micro Filtration (MF)

(10-0.1µm)
Bacteria, suspended particles

Ultrafiltration (UF)
(0.05-0.005µm)
Colloids, macromolecules
Nanofiltration (NF)
5e-3-5.e-4 µm
Sugars, dyes, divalent salts
Reverse Osmosis (RO)
(1.e-4-1e-5 µm)
Monovalent salts, ionic metals

Water

Figure 4: Spiral Wound (SW) Membrane Configuration (Source internet)

7
 Table 1: Comparison of Hollow Fiber and Spiral Wound Membranes
Membrane Advantages Disadvantages
• High membrane surface area to • Sensitive to fouling by colloidal
volume ratio materials
• High recovery in individual RO • Limited number of membrane
Hollow Fiber
unit. materials and manufactures
• Easy to troubleshoot
• Easy to change bundles in the field
• Good resistance to fouling • Moderate membrane surface area
• Easy to clean • Difficult to achieve high recovery
Spiral Wound
• Variety of membrane materials
and manufactures

Table 2. Comparison of Cellulose Acetate and Thin Film Composite Membranes

Parameter Cellulose Acetate (CA) Thin Film Composite (TFC)
Operating pressure (psi) 410 to 600 200 to 500
Operating temperature ( °C) 0 to 30 0 to 45
Operating pH 4 to 6.5 2 to 11
Membrane degradation potential Hydrolyzes at low & high pHs Stable over broad pH range
Permeate flux (gfd) 5 to 18 10 to 205
Salt Rejection (%) 70 to 95 97 to 99
Stability to free chlorine Stable to low (<1 ppm) levels Attacked by low levels(>0.1 ppm)
Resistance to biofouling Relatively high resistance Low resistance
Manufacturer Several Several
Cost Lower 50 to 100 % more

8
 100000 30

Am
qP =15000 gal/day 25
L/A=25 m-1

W
10000

0.028(qP/(qF-qP)k /(1-w S) v
/
5.59 v W (1-w S) PA m

20
20
5000
4000
10000 15 15
3000
2000
10 10
8
1000
6
5

1000 0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
(1 - / P)

Figure 5: Parametric Equation Based Sizing Model