Membranes solve North Sea waterflood sulfate problems

Nov. 25, 1996
Roy Davis Dow Chemical Co. Midland, Mich. Ian Lomax Dow Chemical Co. Dubai, U.A.E. Mark Plummer Marathon Oil Co. Littleton, Colo. To prevent barium sulfate scale from forming in the North Sea Brae field producing wells, Marathon Oil Co. U.K. Ltd. is successfully employing thin-film composite (nanofiltration) membranes for removing sulfate from injected seawater.
Roy Davis
Dow Chemical Co.
Midland, Mich.

Ian Lomax
Dow Chemical Co.
Dubai, U.A.E.

Mark Plummer
Marathon Oil Co.
Littleton, Colo.

To prevent barium sulfate scale from forming in the North Sea Brae field producing wells, Marathon Oil Co. U.K. Ltd. is successfully employing thin-film composite (nanofiltration) membranes for removing sulfate from injected seawater.

In the early 1980s, FilmTec Corp., a Dow Chemical Co. subsidiary, first developed these composite membranes, which now are in their third generation. Marathon Oil Co. holds the patent for the specific nanofiltration membrane process for mitigating scale formation and deleterious reservoir effects.1

This first article in a three-part series describes membrane technology. The remaining articles will detail specific membrane performance characteristics and field experiences in the Brae fields.

Composite membranes

Early in the Brae oil development, Marathon realized that pressure maintenance with seawater injection would be required over the reservoir life. It also discovered that the Brae reservoirs contained significant barium (up to 2,500 mg/l.).

The primary difficulty with seawater injection was that the seawater, with 2,800 ppm sulfate, would react with barium to form barium sulfate scale that would cause flow problems and subsequent plugging in producing wells. A further complication is that barium sulfate scale also contains naturally occurring radioactive material (NORM).

Marathon considered several solutions to the scaling problem, but eventually narrowed these down to:

  • Chemical inhibition

  • Sulfate removal from seawater with composite membranes.

Reference 2 elaborates more on both methodologies.

Marathon worked closely with FilmTec personnel to develop composite membranes that reject selectively sulfate ions in seawater while allowing sodium and chloride ions to pass through the membrane. The result is a product water that is nearly ideal for injection into offshore reservoirs.

After 5 years, the first-generation membranes (FilmTec NF-40) at Brae were replaced by second-generation membranes (FilmTec SR-90).

The second-generation membranes achieved even lower sulfate levels at lower operating pressures, thus reducing power requirements. In addition to minimizing the potential for scale formation and associated well workover and squeeze treatment costs, the reduced sulfate levels are so low that reservoir souring, due to the sulfate conversion to hydrogen sulfide via thermophilic sulfate reducing bacteria, may be controlled.3

In use now is a third-generation membrane (FilmTec SR90-400) that increases membrane surface area by 25%, allowing increased throughputs and thus decreasing space requirements on an oil platform as well as the weight of the sulfate removal facility (SRF).

Nanofiltration membrane

Those conversant with membrane-based liquid separation processes are familiar with reverse osmosis (RO), ultrafiltration, and microfiltration, terms based on the membrane pore size and passage/rejection of specific ions.

FilmTec coined the term nanofiltration to distinguish the pore morphology of newly developed membranes that showed very high divalent ion rejection while allowing monovalent ions to pass. Specifically, a nanofiltration membrane rejects all particles and ions down to 1 nanometer or 0.001 microns (µ).

Marathon's evaluation of the first-generation membrane showed a very high sulfate ion rejection at low pressures, thus decreasing, compared to RO, pumping (energy) requirements. This parameter is important for treating large seawater volumes such as for a waterflood. Fig. 1 [36669 bytes] delineates the generally accepted membrane-based processes.

Membrane construction

The nanofiltration membranes are thin-film composite membranes (also called composite membranes) constructed of several material layers, each providing a specific function.

The membrane surface contacting the feed seawater is a thin film of crosslinked polyamide that provides passage and/or rejection of specific ions that it contacts. This polyamide layer rests on a microporous polysulfone material structure that supports the polyamide layer. These membrane materials reside upon a reinforcing fabric that provides mechanical integrity to the composite membrane (Fig. 2 [15577 bytes]).

The polyamide and polysulfone layers are manufactured separately. The polyamide layer is built for superior water flow and salt rejection while the thicker polysulfone microporous layer is constructed for an optimum in porosity, strength, and resistance to compaction.

Thin-film composite membranes maintain their low transmembrane pressure drop characteristic or low operating pressure requirements for several years. This characteristic was demonstrated by the first set of the first-generation membranes that operated continuously for 5 years at Marathon's Brae complex.

Reference 4 covers further details on nanofiltration membrane composition.

To maximize membrane area into the least space, membranes are constructed in a spiral-wound configuration (Fig. 3 [25848 bytes]). This configuration results in a membrane module that is the key component in an SRF. The spiral wound design contains two membrane layers glued back-to-back on a permeate collector fabric.

The membrane envelope is wrapped around a perforated hollow tube into which the permeate or product water empties from the permeate channel fabric, where it is collected for waterflood injection. A plastic netting is wound into the device on the feedwater side and maintains the feed stream channel height that allows the feed water to be transported across the membrane surface.

This channel spacer also promotes feedstream mixing to minimize concentration polarization or the salt concentration on the membrane surface. Such mixing is designed to eliminate "dead spots" where salts, scale, and other residues can reside and deactivate a portion of the membrane surface.

Two factors that determine what is rejected and what passes through a membrane are pore size and the chemical characteristic of the membrane surface. In nanofiltration membranes for removing sulfate from seawater, the negatively charged polyamide-layer surface causes anion repulsion and determines salt rejection selectivity.

For example, calcium chloride rejection, with the chloride ion as the monovalent anion, is about the same as that of sodium chloride. However, divalent and multivalent ions such as sodium sulfate or sodium phosphate are rejected because of their greater anionic charge.

With larger ionic salts, pore size also becomes increasingly important and accentuates the membrane's chemical characteristic in determining salt rejection. Notably, to maintain electronic neutrality, for every anion, one cation must be removed. Consequently, if one sulfate ion is removed, two monovalent sodium ions or one divalent magnesium ion are also removed.

With the first-generation nanofiltration membrane, the rejection of monovalent ions decreases (passage increases) with increasing salt concentration. The Donnan exclusion principle explains this reaction. At higher cation concentrations, more cations are available to shield the negative charges on the membrane, making it easier especially for monovalent anions to pass through the membrane.

Fig. 4 [19023 bytes] compares rejection of sodium chloride and magnesium sulfate for the first-generation membrane. At 2.0 MPa (290 psi) pressure, sodium chloride rejection drops from 60% at 0.2% sodium chloride solution to 20% at 4.0% sodium chloride solution.

Note that a minor part of this rejection loss is a function of the 35% decrease in permeate flux, but the major part is due to a five-fold increase in the salt permeability through the membrane. However, the sulfate ion charge density is high enough to be almost completely rejected by the membrane, even at a high, 40,000 mg/l., ionic strength magnesium sulfate solution.

This high sulfate ion charge density simply overwhelms the ability of the sodium ion to shield it from the negative charge of the membrane. This chemical characteristic combined with the one-third greater size (2.4 Å. for the sulfate ion-vs.-1.8 Å. for the chloride ion), simply does not allow it to pass through the membrane layer.

Note also that by allowing sodium and chloride ions to pass through the membrane while rejecting sulfate ions, one does not need to overcome as high an osmotic pressure across the membrane surface. As a result, operations can be at lower pressures that reduce operating costs.

Secondly, by leaving sodium and chloride ions in the product water, one avoids downhole clay swelling in the reservoir.

Membrane performance

Because the first-generation membranes selectively reject the sulfate ion while allowing the sodium and chloride ion to pass through, they fit Marathon's needs at Brae. A 40,000 bw/d SRF was installed and commissioned on the South Brae platform in November 1988. Membrane performance was better than predicted by an earlier offshore pilot plant operation. The SRF met all technical and economic criteria established by Marathon.

Once the full-scale system was in operation, areas for potential improvements were identified. The first potential area for improvement was the sensitivity to temperature.

It is well known, that at lower temperatures, feed water viscosity will increase. To overcome this viscosity increase, higher pressures are needed to diffuse the water through the membrane.

Consequently, in a waterflood, to maintain the needed volume of low sulfate water, the feed pressure is increased to compensate for increased membrane resistance. Usually, the required feed pressure increase is 3%/°C. decrease in feed water temperature. However, at the lower feed water temperatures in the North Sea, an increase in the percent chloride ion rejected also occurred.

The net result was a higher chloride ion concentration on the feed side of the membrane compared to the product or permeate side. This increased osmotic pressure across the membrane and required additional pressure difference to maintain an equilibrium. This resulted in a required feed pressure of 4.8%/°C., instead of 3%/°C.

Based on the pressure increase required to maintain the waterflood volume requirements at temperatures less than 15° C. (59° F.), the 600 psi operating pressure exceeded the structural pressure limits of the spiral wound membrane module. The Brae platform was faced with either increasing feed water temperature or dropping back to the maximum module operating pressures, thus decreasing the SRF output.

Initially, Marathon selected the option of increasing the feed water temperature using waste heat.5

As a result, FilmTec initiated a research program focused on altering/modifying the physical (porosity) and chemical (charge) characteristics of the polyamide membrane layer. By establishing a more thorough fundamental understanding of each membrane component and process, FilmTec designed a second-generation nanofiltration membrane.4

The second-generation membrane exhibited reduced chloride ion rejection (more chloride was allowed to pass) and a higher flux or output rate, two characteristics that are generally consistent with the membrane porosity structure.

Additionally, surface modifications allowed the polyamide layer to differentiate more precisely between the charge densities of the monovalent (chloride ion) and divalent (sulfate ion) anions. This reaction enhanced the ability of the polyamide layer to reject the sulfate ion while allowing an increased chloride ion concentration to pass through the membrane.

Table 1 [6881 bytes] shows a typical analysis of seawater treated with the first and second-generation membranes. The accompanying box summarizes tests on the two membranes.

Membrane upgrade

To reduce capital costs, weight, and footprint size of an SRF, important for offshore platforms, FilmTec developed a third-generation membrane module.

In the same diameter configuration, this module contains 25% higher membrane surface area than either the first or second-generation.

Two offshore platforms now use third-generation membrane modules.

SRF maintenance

A reliable and cost-effective membrane system depends upon keeping membranes clean and free of foulants. Foulants on the membrane surface prevent a system from providing the needed output and sulfate removal characteristics as well as the membrane life necessary for economic viability.

Pretreatment measures and operational discipline are needed to minimize membrane fouling that results in low maintenance and, therefore, the reliable SRF performance.

Particle filtration

As discussed earlier, the membranes are rolled into a spiral wound configuration to maximize the membrane surface area in the least space. In this spiral wound configuration, a spacer (plastic netting) is positioned on the membrane surface (Fig. 3 [25848 bytes]) so that the feed water can flow across the membrane surface and from the inlet to the outlet of the membrane module. By removing particles in the feed water that can be wedged or deposited where this spacer comes into contact with the membrane surface, one can maintain the original active surface area and membrane performance.

This particle removal is accomplished by the addition of a polyelectrolyte that coagulates particles to a sufficient size so that they can be removed via media filtration. Media filtration is generally accomplished with multimedia filters that remove all particles greater than 5 µ. Choosing the correct polyelectrolyte and dosage rate ensures the optimum media filter operational efficiency.

Cartridge filters downstream of the multimedia filters provide a back-up to the multimedia filters and an "insurance policy" for the membrane modules. These cartridge filters that can be selected to remove all particles down to 0.5 µ, also safeguard the membrane modules from carryover of sand or corrosion products from the multimedia filter.

Although 0.5 µ nominal-rated filters are used, 5 µ nominal rated with 30-40 absolute are generally sufficient. Both these pretreatment steps are economically sound and essential to a good SRF design.

Bacteria remediation

The close voids within the spiral wound membrane module could provide a surface for bacteria growth and subsequent biofilm formation resulting in biological fouling. Such potential for biological fouling is prevented by adding sodium hypochlorite to the feed water. The sodium hypochlorite is generally produced by a "hypochlorite generator" that uses the sodium chloride in the seawater for the source of the hypochlorite or free chlorine. This free chlorine destroys the bacteria content present in the incoming feed seawater, thus preventing a problem before it occurs.

The subsequent addition of sodium metabisulfite, at 10% to 30% above the stoichiometric level, assures complete removal of chlorine that could oxidize the polyamide membrane. This combined with the placement of the deoxygenation system or deaerator, for the injected water, upstream of the second-generation membranes results in a seawater feed with an oxygen content in the parts per billion range. This further reduces the biofouling potential due to the lack of oxygen for aerobic bacteria.

Operations

Generally, six membrane modules are placed in each vessel, with vessels arranged in a 2:1 array. This array allows for maximum pumping (energy) efficiency and a recovery of 75% of the feed water for injecting into the reservoir.

By "funneling" rejected water from each of two vessels in the first array to the feed stream for the second vessel, sufficient cross flow velocity is maintained to ensure optimum feed water contact with the membrane surface.

Because of the 2:1 array (or arrangements of the vessels with a 2:1 ratio), the feed water will pass through 12 membrane modules prior to discharge:

2 arrays x 1 vessel/array x 6 modules/vessel = 12 membrane modules

During this time, the sulfate concentration, for example, will increase from 2,800 mg/l. to over 11,000 mg/l. as it is being rejected or repelled from the membrane surface.

Because one-third of the calcium is also being rejected from the membrane surface, simultaneous increase in the concentration of both these species could cause membrane fouling from calcium sulfate precipitation on the concentrate side of the membrane.

However, antiscalants will prevent this precipitation. Such antiscalants, generally added in the 5 ppm range, function by retarding calcium sulfate precipitation until the rejected calcium and sulfate ions are discharged. Various effective commercial antiscalants are available.

Chemical cleaning

To maximize membrane module life, typically 4-5 years, chemical cleaning at a certain interval is recommended.

Adequate pretreatment steps and antiscalants will reduce chemical cleaning frequency. Chemicals will clean colloidal foulants and insoluble organic constituents that build on the membrane surface.

The membranes can be cleaned in a 3-10 pH range and at up to 35° C. (95° F.). Iron and calcium salts can usually be removed at a 3 pH with hydrochloric or citric acid, whereas biofilms and organics require a dilute solution of sodium EDTA or detergent at a 10 pH.

The cleaning solution is made up in a separate tank and circulated through the membranes when the system is shut down. To reduce the volume and size of the cleaning system, only one bank or part of the system is cleaned at any one time.

Cleaning also requires a pump and a cartridge filter for removing any undissolved chemical reagent. Water for the cleaning solution is generally from the platform's potable water system.

The chemical cleaning tank can also be used to make up membrane storage solutions of glycerin/sodium metabisulfite and nonoxidizing biocide solutions of formaldehyde/glutaraldehyde for recirculation. Alternatively, formal dehyde/glutaraldehyde can be used in periodic shock doses to the membrane feed as an additional precautionary step to prevent biological fouling.

References

1. U.S. Patent 4,723,603.

2. Hardy, J.A., and Barthorpe, R.T., Plummer, M.A., and Rhudy, J.S., "Control of Scaling in the South Brae Field," Paper No. OTC 7058, Offshore Technology Conference, Houston, 1992.

3. Eden, Robert, et al., "Oil field Reservoir Souring," Offshore Technology Report 92 385, University of Manchester Institute of Science & Technology, Capcis, 1993.

4. U.S. Patents 4,769,148 and 4,859,384.

5. Weir Westgarth Ltd., "Sulfate removal from injection water," World Expro 93.

The Authors

Roy A. Davis is a development leader, liquid separation technology, for Dow Chemical Co. in Midland, Mich. He has been with Dow since 1971 and has considerable experience in a wide variety of water treatment processes, but most recently in membrane technology to treat water in offshore petroleum operations. Davis has a BS from Central Michigan University.
Ian Lomax is a development manager for Dow Chemical Co. and is presently assigned to the liquid separations technology's operations in Dubai, U.A.E. In 1985 he joined FilmTec Corp. a subsidiary of Dow, and was involved in the design of Marathon Oil Co.'s offshore sulfate removal system. Lomax is registered as a Chartered Engineer and member of the Institution of Chemical Engineers in the U.K.
Mark Plummer is an advanced senior engineer with Marathon Oil Co. in Littleton, Colo. He is currently involved in process development for decomposing hydrogen sulfide, and in optimizing conditions for manufacturing asphalts under new specifications. Plummer is a graduate of the Colorado School of Mines, Golden, Colo.

Tests compare first and second generation membrane

Steven J. Tulloch
Environmental & Resource Technology, Ltd.
Orkney Water Test Centre Flotta, Stromness, Orkney Isles

INDEPENDENT TESTS ON THE FIRST AND SECOND-GENERATION membranes at the Orkney Water Test Centre (OWTC) in the Orkney Isles, U.K., used a 1 cu m/hr nanofiltration plant manufactured by Salt Separation Services, utilizing Dow membranes.

The evaluation compared, side by side, long-term fouling resistance of both generations of membranes. Also investigated were the effects of varying flux rate and temperatures, with respect to operational performance and ion removal efficiency.

The test used seawater taken from the Scapa flow in Orkney. This water is very similar in most physical and chemical characteristics to seawater in the central North Sea. Both generation of membranes were rolled into spiral wound modules with other components.

The membrane test unit had two parallel trains or systems of two vessels in series followed by a single train of two further vessels in series (each vessel contained two membrane modules).

For both membranes, trial results are summarized as follows:

  • Both generations of membranes, NF40 and SR90, exhibited high sulfate rejection (removal) and low chloride rejection (SO4 98% rejection, Cl < 8%).

  • The transmembrane pressure drop (pressure across the membrane surface required to maintain a constant output) increased gradually with time due to fouling. This was easily corrected by cleaning approximately every 3 weeks.

  • Sulfate rejection improved when the membranes were operated at a higher flux rate although the rate of fouling also increased at the higher flux rate.

  • Sulfate rejection decreased with increasing seawater temperature (about 1 ppm SO4/°C. in the 15-20° C. range).

  • Differences between the membrane generations can be summarized as follows:

  • Second-generation membranes improved sulfate rejection compared to the first-generation membranes (99% vs. 98%).

  • Second-generation membranes lowered chloride rejection more than the first-generation membranes (4% vs. 9%.)

  • Second-generation membranes operated at a lower transmembrane pressure drop (3%/°C. vs. 4.8%/°C.) than the first-generation membranes (partly as a result of the lower chloride rejection).

These results were consistent with earlier studies on the first-generation membrane by Torleiv Bilstad1 (Rogaland University Centre, Norway).

The trials, at OWTC, concluded that the second-generation membrane improved sulfate ion removal performance and improved selectivity for removing sulfate ion-vs.-chloride ions. Consequently, the second-generation membranes require lower feed pressures, and membrane output is improved at the lower temperatures characteristic of the North Sea.

Acknowledgments

Sponsors for the work at OWTC were Elf Aquitaine, Hamilton Oil Co., Marathon Oil Company U.K., Norsk Hydro a.s., Phillips Petroleum Co. Norway, Saga Petroleum a.s., Shell International, Petroleum Maatschappij, and Den norske stats oljeselskap a.s. (Statoil). Their contribution to this evaluation is gratefully acknowledged.

Reference

1. Bilstad, Torleiv "Sulphate Separation from Seawater by Nanofiltration," International Produced Water Symposium, San Diego, February 1992.

Copyright 1996 Oil & Gas Journal. All Rights Reserved.