Keith O'DonnellThe world's first sulfate removal facility (SRF) on the Brae A production platform in the central North Sea demonstrates the effectiveness of membrane technology with only a few minor problems caused by the retrofit nature of the installation.
Marathon Oil UK Ltd.
Aberdeen
This is the second in a three-part series that started in OGJ, Nov. 25, p. 59. It details experiences with membrane technology on the Brae A platform that future users of this membrane technology can use for optimizing their SRF installations.
Barium sulfate
Formation water in the south and central Brae reservoirs contains very high levels of barium ions. Consequently, there is a high potential for forming barium sulfate scale when Brae formation water is mixed with seawater.
Because of high levels of barium, conventional methods for preventing barium sulfate scale with chemical scale inhibitors proved difficult and expensive, and are of limited value for protecting the reservoir matrix.
Injecting potable-grade water into the reservoir also was not a solution. Although this would prevent barium sulfate scale, it potentially could damage the formation by causing clays in the reservoir matrix to swell and block pores.
Therefore, the Brae field required a process that could selectively remove sulfate ions from seawater yet retain most other salt components. Reverse osmosis appeared to be one option, and subsequent collaboration with FilmTec identified a membrane that would only pass particles of 1 x 10-9 m (nanofiltration) and smaller.
This membrane permitted passage of most sodium and chloride ions but let only a small percentage of sulfate ions through.
SRF development
A membrane pilot plant installed on the Brae A production platform determined operating parameters for designing a sulfate removal facility (SRF) capable of providing sufficient quantities and quality of low-sulfate water for waterflooding the reservoir.
Following pilot plant trials in 1987, a full-scale SRF was installed on Brae A in three phases.
Train A of the Brae A SRF was commissioned in December 1988, followed by Train B in August 1989 and Train C in March 1990. All three trains were fitted with FilmTec's first-generation sulfate removal membranes.
Each train was nominally rated for 40,000 b/d of low-sulfate seawater (LSSW) when operating at a 75% recovery rate (Volume LSSW/Volume feed seawater). The total (nominal) installed LSSW capacity was 120,000 b/d.
Train B was retrofitted with the second-generation of sulfate removal membranes in March 1993. Train C was fitted with the third-generation membranes in April 1995. After more than 5 years of operation, in July 1995, the original first-generation membranes in Train A were replaced with the third-generation membranes.
The second and third-generation membranes are both less temperature dependent than the original first-generation membranes, allowing greater throughput, and providing increased sulfate removal.
Plant description
Fig. 1 [65111 bytes] is an overview of the Brae A SRF. Deoxygenated seawater is pressured to 20-32 bar (290-464 psi) by the SRF feed pumps. The water is then filtered in the SRF feed inlet cartridge filters to 5 microns () nominal before being piped to the three trains or independent membrane systems.
The three SRF trains are identical. Each train consists of four parallel banks of vessels. Each bank is rated at 10,000 b/d of low-sulfate seawater. The size of these banks was based on the maximum size of the unit or skid that could be maneuvered and placed in the tight confines of the module support frame of the platform.
Each bank consists of 15 vessels, each vessel containing six membrane elements. The 15 vessels are arranged as 10 parallel "front" vessels and 5 parallel "rear" vessels (2:1 array). Water is fed to the 10 "front" vessels, and an LSSW product and a brine stream leave the vessel in a 1:1 ratio.
The LSSW product flows to the product header. The combined brine stream from the 10 "front" vessels is then fed to the 5 "rear" vessels. An LSSW product leaves these vessels and joins the product water from the 10 "front" vessels in the product header.
A concentrated brine stream leaves the "rear" 5 vessels in a 1:1 ratio with the LSSW product and is routed to the seawater disposal caisson (Fig. 2 [121526 bytes]).
The LSSW product stream is combined with the product streams from the other two SRF trains in the LSSW product header. From this header the water is pressured to the injection pressure via a booster pump and two parallel injection pumps.
Operations
Pilot plant trials had established that the membrane process could satisfactorily operate on a production platform. The full-scale plant design was based on experience of the pilot plant operation and conventional potable water reverse osmosis plants. However, some operational difficulties were caused by limitations imposed because of the project's retrofit nature.
One limitation, as previously described, was the modular design needed to retrofit the installation into an existing operational production platform.
Another limitation was that an existing water injection system had to provide plant feed. This system required deoxygenated seawater to be piped through existing carbon steel pipe in the deoxygenated part of the system and feedwater to be preheated by the platform's cooling medium system. This was fortuitous and not by design.
Also, a backpressure had to be maintained on the product water stream to forward it to the existing water-injection system. This was further complicated by a local control system physically distant from the existing central water injection system controls.
Deoxygenated seawater
The combination of deoxygenated seawater and carbon steel pipe, both resulting from the retrofitting of an existing seawater injection system, introduced problems of biofouling, prefilter blockages, and cleaning.
Locating the SRF downstream of the deoxygenation unit did have the advantage of protecting the membranes against damage from the chlorine that is dosed into the inlet stream of the seawater lift pumps for the control of marine growth. The deoxygenation process removes chlorine as well as oxygen from the feed seawater.
Early operation of Train A did not include a biocide to control biological growth within the membrane element. However, biocides were added to the original high-sulfate seawater for the normal control of downhole bacterial growth. The initial hesitation to use biocides in the SRF was due to concerns about chemical damage to the first-generation membranes with the potential permanent loss of capacity and/or reduction in sulfate ion removal efficiency.
The feed pipe to the SRF was uncoated carbon steel. In the absence of biocides, sulfate-reducing bacteria (SRB) became established causing a layer of microbiological debris and corrosion products to form on the inside of the pipe. This prematurely blocked the SRF feed inlet cartridge filters and deposited solids within the leading membranes of the "front" 10 vessels in each of the four banks of Train A.
The situation was corrected by:
- Acid cleaning of the feed pipe to Train A to remove the biofouling debris.
- Replacing the leading membranes in the "front" 10 vessels of each of the four banks in Train A
- Establishing a surfactant-free biocide regime that treated the whole SRF and associated feed pipe including the deaerator.
Cleaning the fouled membranes was possible but was not economically viable.
A biocide, an aldehyde-based product, was selected following discussions with FilmTec and laboratory testing that confirmed its compatibility with the first-generation membrane.
The success of this biocide has been confirmed in day-to-day operations and it now is being used with the second and third-generation membranes.
Temperature effects
Output from the first-generation membrane is temperature dependent. Consequently, in the early SRF operations, a maximized seawater feed temperature was needed to obtain the required LSSW output.
This maximization used seawater previously employed as a coolant in the closed-loop cooling medium system for various processes. Maximized SRF feed temperature was achieved by balancing flows around and through the coolers.
In the early months of operation, feed temperatures were 25° C. (77° F.). This was the maximum temperature achievable without limiting compression in the Brae A gas processing plant.
In August 1990, modifications to the Brae A gas processing plant included removal of two gas compressors. The effect of removing these heat sources reduced the maximum achievable temperature to 19° C. (66° F.). This change reduced plant capacity by 18% and prevented the SRF from operating at its original 75% recovery (Fig. 3 [103296 bytes]).
To maintain SRF at 19° C. (66° F.), it was necessary to adjust the operating parameters. It was not considered prudent to increase feed pressure because experience showed that this would significantly increase the rate of membrane fouling and membrane cleaning frequency.
Thus, low-temperature operations required reduced recovery from the SRF. Discussions with FilmTec developed an operating envelope (Fig. 3 [103296 bytes].) In this way, the optimum brine flow could be maintained albeit at the expense of the LSSW product rate.
With this change, plant operations at low feed temperatures could be maintained within the devised operating envelope without incurring excessive membrane fouling. The net effect reduced recovery from 75% to about 55%, which at that time had little effect on reservoir management.
The feed temperature effects on the first-generation membrane capacity prompted FilmTec to develop the second-generation membranes. Because the second-generation membranes are less temperature dependent, the operating recovery has been returned to a set 75% on Train B of the Brae A SRF.
A further benefit of the second-generation membranes has been that the SRF plant operates at the full 75% recovery even when the hydrocarbon production plant is shut down and SRF feed temperatures are a low 6° C. (43° F.). The first-generation membranes did not have this capability. The new third-generation membranes also exhibit these low temperature benefits.
The first-generation membranes, despite the temperature effect problems, gave good service in the early years of the Brae A SRF. The membranes had a 3.5-5 year operating life (depending on which phase they were fitted in) and produced an LSSW product in sufficient quantities and quality to satisfy Brae field needs.
Originally, first-generation membrane operating life was expected to be only 3 years.
Control System
The SRF control scheme was more complex than operational experience would suggest was necessary.
The control scheme permitted operating three parallel trains and fulfilled the requirement of maintaining a backpressure on the product water stream so that it would flow to the existing water injection system without booster pumps.
This scheme is contrary to the normal practice in reverse osmosis systems where there is an ambient pressure break tank on the product side of the membranes. The backpressure requirement complicated instrumentation for protecting the membranes from reverse flow under upset (trip) conditions. The absence of a break tank made the downstream pumps liable to trip when small flow fluctuations occurred.
The control scheme was further complicated by being remotely located relative to the SRF as well as from the existing centralized water-injection system controls.
The SRF requires three production operators for start-up. Plant operation has proved to be stable and, as such, the automatic flow controls are routinely placed in manual override. This change has allowed the original control complexity to be satisfactorily bypassed
SRF optimization
An area where improvement might be possible is improved chemical treatment/water pretreatment to reduce the gradual fouling which requires units to be periodically cleaned. This effort needs to be further justified against current operating cost, which although not excessive, might not be optimum.
A subsequent SRF, installed on the Ewing Banks platform in the Gulf of Mexico, includes this optimization. Membranes there have been in use nearly a year with little transmembrane pressure drop increase and less frequent cleaning.
SRF performance
Despite the prototype nature of the Brae A SRF, it has run satisfactorily for nearly 7 years and produced over 90 million bbl of LSSW for injection into the South and Central Brae reservoirs.
At 2-4 pence/bbl ($0.03-0.06/bbl) LSSW costs typically six times more than conventional seawater injection. Consequently there is a continuing need for minimizing SRF operating costs.
Operator time could be further minimized by centralizing and integrating the SRF control scheme with that of the water injection system and keeping it as simple as possible.
In a newly built plant, the number of independent banks or membrane systems should be minimized. For the Brae A plant, platform space limited the sizes of the banks to 10,000 bw/d. In a new facility this should not be a limitation.
Designing an SRF as part of a waterflood treatment system on a new platform should also allow plant controls to be located at the SRF plant.
Controlling membrane scaling/fouling at Brae A would benefit from applying the newer approaches utilized at Ewing Bank.
Such investigation should include examining the corresponding best method of membrane cleaning. This includes design of the wash equipment and selection of the most suitable cleaning chemical that will minimize cleaning time and cost.
The Author
Keith O'Donnell is an operations manager with Brown & Root Ltd. in Aberdeen, Scotland. Previously, he held various positions with Marathon Oil U.K. Ltd. The last was platform manager on the Brae A platform.
O'Donnell has a BS in chemical engineering and a PhD from the University of Aston in Birmingham, England. He is a Fellow of the Institution of Chemical Engineers and a member of SPE.
Copyright 1996 Oil & Gas Journal. All Rights Reserved.