Sea water reverse osmosis - energy efficiency & recovery

Recognising that sea water reverse osmosis (SWRO) requires significant amounts of energy to produce drinking water (relative to conventional treatment), desalination plant designers carefully assess the energy implications early in the design phase.

The extent of the energy efficiency initiatives and how they were analysed, evaluated and applied is what distinguishes the Adelaide desalination plant from other desalination plants. For example, the project team operated a pilot plant for over 12 months prior to the project commencing and used this facility to optimise the water treatment process and evaluate alternative processes and technologies.

The paper presents the process employed at the Adelaide desalination plant, considers certain aspects of the design in terms of energy efficiency and recovery and discloses whether those design parameters have been achieved in terms of the operational performance of the plant.

The Adelaide desalination plant is designed and constructed in two separable portions (SPs) each with a maximum water production capacity of 150 ML/day creating a total plant capacity of 300 ML/day. The plant is an excellent example of high efficiency, durability, flexibility with high levels of availability. Efficient energy use throughout the intake, pre-treatment, reverse osmosis (RO), post-treatment and washwater systems provides the plant operator with specific energy consumption below the required 4.1 kWh/m³ for all conditions experienced thus far within the design envelope (feed salinity minimum of 36,000 mg/L @ 12C and maximum of 42,000 mg/L @ 26C). The minimum design life requirements of 25 years for process assets, 50 years for structures and 100 years for civil assets (including the intake and outfall tunnels) provide the plant operator with high levels of durability whilst minimising asset maintenance costs. Flexibility in production rates from 30 ML/day up to 300 ML/day in 15 ML/day increments allows the South Australian Water Corporation to optimise supply to accommodate consumer demand and network conditions. Plant reliability, availability, operability and maintainability were considered from the commencement of the design phase, with the operator providing valuable input during design, construction and commissioning phases of the project. Failure mode, effect and criticality analysis modelling undertaken on the Adelaide desalination plant predicts an operational availability of greater than 96% (Arito et al. 2013) (Figure 1).

The process objective of the intake system is to draw seawater through the intake tunnel, screen the seawater to remove particulate matter and transfer the seawater to the pre-treatment system. The open seawater intake system is located approximately 1.3 km offshore, the high point of the intake riser is located about 12 m below sea level and the intake tunnel starts about 34 m below the water's surface. Chemical dosing of sodium hypochlorite and sulphuric acid close to the seawater intake point maintains pipeline hygiene and minimises marine growth on the internal tunnel wall. The intake pumping station is housed within a spectacular, large in-land cavern located in the coastal zone 8 m below sea level. An up-surge chamber mitigates flood risk to the cavern in the event of a pump station trip when running at full capacity under extreme high sea level conditions. Within the cavern, travelling band screens remove particulate matter greater than 3 mm consisting of seaweed, crustaceans and other debris from the seawater. The intake pumping system comprises 12 large pumps which transfer the screened seawater to the disc filters and then into the open feed channel of the ultrafiltration (UF) system which is located 52 m above sea level in the main plant.

The electrical energy required by the intake pumps to transfer 600 ML/day of seawater from sea level to the main plant (located 52 m above sea level) represents 16% of the total energy consumed by the plant. Measurements taken during the recent annual capacity test at Table 1 show that for a plant output of 300 ML/day the operator is achieving an additional energy saving of 5% compared to that predicted by the design. This improvement reduces the specific energy consumption of the overall plant by 0.03 kWh/m³, (or 1.0%).

Table 2 compares the electrical energy required by the intake pumps to lift 600 ML/day of seawater from sea level to the main plant (located 52 m above sea level) with the theoretical minimum energy usage (potential energy required to lift the water assuming no friction). The measured 90% intake system efficiency is favourable given that water lifting systems typically achieve between 50 and 90% efficiency depending upon their system configuration and requirements.

Seawater filtering prior to UF is required to remove particles greater than 100 microns that have passed through the intake pumps and have not been removed by intake band screens. This filtering is undertaken using two parallel trains of disc filters, each train has 6 arrays, and each array has 12 disc filters. The disc filters are required to protect the UF system from fine sand, shells and some micro-organisms (such as barnacle larvae) which have sharp edges that may damage the membrane fibres.

The filtered raw water from each train is then fed to the associated UF channel. A submerged type of UF system has been installed to treat water to a high standard suitable for feeding into the RO membranes. The filters are able to filter out microscopic particles including bacteria and viruses. The silt density index of the filtrate is monitored to verify that the filtrate is of the required quality. This UF system consist of two feed channels each feeding seven membrane cells. Each SP of the plant therefore contains a total of fourteen open UF cells and fourteen filtrate pumps. Water is gravity fed to the cells from the feed channel with the water level in the channel and the cells fixed by the hydraulics of the feed channel inlet and the control of the intake pumps.

The UF membranes are subjected to regular backwash cycles carried out at intervals of 20–90 minutes and lasting approximately 3 minutes. To ensure filtrate flow is maintained, no cell is taken off-line for regular backwash until one of the cells in stand-by has been put into operation and has reached the steady flow production conditions. In addition to this, periodic chemical cleaning-in-place (CIP) maintains performance of the UF system. The UF system operates continuously 24 hours a day, 7 days a week as required without reducing output during backwash and CIP. UF filtrate is then fed into the cartridge filters prior to the RO system.

A key feature of the ADP process design was to directly couple the UF and the RO systems without the need for an intermediate tank. This creates the situation where all the surplus energy in the UF filtrate flow is transferred to the RO system, hence reducing the energy required in the RO pumps. The estimated energy saved through not having intermediate tanks reduces the specific energy consumption of the overall plant by 0.07 kWh/m³, (or 1.8%).

Cartridge filters are located between the UF and RO systems before passing this filtrate to RO membranes. The primary function of the cartridge filters is to protect the sensitive RO membranes in the event of significant failure of the upstream UF system.

The Adelaide desalination plant's RO process is divided into two parallel trains. Figure 2 depicts the general arrangement of the trains within each SP. Each train comprises 5 first pass racks [A1–A5 in train A, and, B1–B5 in train B]. The first pass racks A1–B5 also incorporate a partial second pass (termed second pass front) where permeate undergoes further treatment. The remainder of the first pass permeate is processed through separate second pass rear RO racks [C1–C5] located at the far end of the RO hall; with racks C1–C3 aligning with first pass Train A and C4–C5 with first pass Train B.

The Adelaide RO configuration consists of a non-conventional two pass array designed to optimise flux distribution through the first pass, which seeks to reduce potential fouling and extend membrane life. The configuration uses a blind split between the 3rd and 4th membrane elements resulting in a more even flux decline through the first pass (Hijos et al. 2011). This arrangement is designed to lower the membrane replacement rate with only a small penalty in energy consumption compared to a conventional design. A 50% recovery is achieved from the 1st pass membranes and 90% recovery from the 2nd pass membranes as shown in Figure 3. The Adelaide RO system achieves an overall recovery of 48.5%.

The high-pressure pumping system (RO feed booster pump and RO high-pressure pump) and the energy recovery system (pressure exchangers and booster pump) feed pre-treated seawater at high pressure to the 1st Pass RO. Permeate extracted through the front 3 membranes of the 1st Pass is processed directly by the 2nd Pass front RO where salinity is further reduced without the need for further pumping.

A key feature of the RO system is that permeate from the rear 5 membranes of the 1st Pass is blended with reject from the 2nd Pass front RO to be processed through the 2nd Pass rear racks. The 2nd Pass rear booster pumps control the flow and pressure to these rear racks as required. Permeate from the 2nd Pass front and rear RO is blended in the common permeate header and is directed to the treated water storage tank and the post-treatment process. The 2nd Pass rear RO reject is re-circulated to the head of the RO system slightly diluting the feed to the 1st Pass. The saline concentrate from the 1st Pass RO is discharged back into the sea after recovery of waste energy.

The pressure exchanger energy recovery devices which are employed in the RO system reduce the specific energy consumption of the RO system by more than 42%. The devices use the principle of positive displacement and isobaric chambers to achieve extremely efficient transfer of energy from high-pressure brine stream (the saline concentrate stream), to a low pressure incoming feed stream (the seawater stream). Almost no energy is lost in the transfer which uses positive displacement to pressurise UF filtrate by direct contact with the high-pressure saline concentrate stream from the 1st Pass RO.

The electrical energy required by the RO system represents a very significant 80% of the total energy consumed by the plant. The design incorporates energy saving measures such as removal of the intermediate tanks between the UF and RO systems, introduction of highly efficient RO energy recovery devices, the use of a combination of fixed and variable speed pumps to maximise efficiency and the use of non-conventional two pass array RO design. As a result of these and other efficiency and energy recovery techniques, operation of the plant shows at Table 3 that for an output of 300 ML/day the operator is achieving an additional energy saving of 3% from that predicted in the design.

Table 4 shows that the operator is achieving an additional specific energy saving of 3% from that predicted in the RO design. This reduces the specific energy consumption of the overall plant by a further 0.09 kWh/m³, (or 2.4%).

Post-treatment is essential for the RO permeate to ensure that it is fit for drinking purposes. Post-treatment includes remineralisation, conditioning and disinfection. Remineralisation is carried out by dosing carbon dioxide and limewater. Conditioning uses hydrofluorosilicic acid for fluoridation. Disinfection uses gaseous chlorine to prevent bacteria growth during storage and transport of treated water. Treated water is transferred to two 25 ML drinking water storage tanks in preparation to transfer via the transfer pumping station pipeline to the South Australian Water Corporation drinking water network.

The elevation of the process plant at 52 metres above sea level makes it economically viable to treat and recycle dirty washwater streams from the pre-treatment system. This involves the inclusion of a washwater treatment facility close to the process buildings that treats the backwash water generated by the disc filters and the UF system. The clarified washwater from this treatment system can then be recycled (via gravity flow) to the head of the UF channels. This aspect of the design allows the pre-treatment recovery to approach 100% and has the potential to reduce the specific energy consumption of the overall plant by 0.015 kWh/m³ (or 0.3%).

The elevation of the process plant at 52 metres above sea level also makes it economically viable to recover some of the potential energy of the saline concentrate before it is discharged to the marine environment. The plant employs 2 Francis turbines for this function, with the electrical power generated from this flow assisting with the electrical demand of the intake pump systems. The design predicts that the energy recovered from the saline concentrate reduces the overall plant energy consumption by 0.10 kWh/m³ (or 2.5%) (Table 5).

Table 6 shows that the actual electrical energy recovered from the saline concentrate flow represents 86% of the theoretical maximum potential energy available (assuming no friction losses). This figure is just below the bench mark of 90% for similar systems.

Buildings design and solar panels installed on a number of the buildings contribute to a more environmentally sustainable plant. Each RO building includes a 100 kW cell array providing a site capacity of 200 kW at peak sun hours. These solar panels provide power to many of the ancillary non-process related functions on site.

The Adelaide desalination plant employs a variety of innovations and technologies in its design to minimise energy use and hence to reduce energy costs. In addition to this, at an early stage in its operational life the plant has achieved an increase in efficiency beyond that anticipated in the design. It must be acknowledged that energy consumption is likely to increase as the assets and membranes age. The factors included in the design which contribute significantly towards making the Adelaide desalination plant one of the most efficient in the World includes the removal of intermediate tanks between the UF and RO systems, the use of a non-conventional RO configuration which creates a higher overall recovery, the use of RO energy recovery devices, recovery of energy on the saline discharge stream, high recovery of the UF pre-treatment system, potential for reuse of washwater from the disc filters and UF backwash and the solar panels installed on a number of the buildings.

Further to this, the plant has experienced improved operational efficiencies beyond those anticipated in the design in the areas of the intake pumping station and RO pumping system which represents 16% and 80%, respectively, of the overall energy consumed by the process, as summarised in Table 7.

For comparison purposes the predicted design values shown in Tables 7 and 8 have been calculated based on a water production of 300 ML/day, seawater salinity 36,540 mg/L, seawater temperature 21.2 °C and RO membrane age of 1.95 years.

Table 8 shows the specific plant power demand and energy consumption and is based on measured results recorded during the plant's recent annual capacity test conducted during the period 3–9 March 2014.

Operational records for 9 March 2014 are as follows: water production of 300 ML/day, seawater salinity 36,540 mg/L, seawater temperature 21.2 °C and RO membrane age of 1.95 years.

These operational performance figures demonstrate that the specific energy consumption of (3.48 kWh/m³) measured during the annual capacity test is a 3.6% improvement on the value predicted during the design (3.61 kWh/m³) and an 8.4% improvement above the initial requirements established at the onset of the project (3.80 kWh/m3).

Survey data within Australia shows that the specific energy consumption for seawater RO plants is typically 3–3.7 kWh/m³ [5] with the current established benchmark being 3.5 kWh/m³. The Adelaide desalination plant with a specific energy consumption of 3.48 kWh/m³ aligns well with the benchmark figure even though the main plant is located 52 m above sea level.

Recognising the high energy intensity of SWRO treatment compared to conventional drinking water treatment processes and the requirements of the South Australian Water Corporation, the AdelaideAqua design team ensured a strong focus on optimisation of energy consumption.

The Adelaide desalination plant incorporates energy efficiency initiatives not seen in other large-scale municipal desalination plants, including direct coupling of UF pre-treatment and RO membrane trains, washwater treatment and recycling capability, outfall energy recovery and photovoltaic energy generation from solar panels. Outcomes of a recent plant operational performance test demonstrated an energy saving of 8% when compared to the project's initial design requirements.

In summary, the Adelaide desalination plant's process design enables a high degree of energy efficiency, whilst meeting all client requirements relating to production capacity, drinking water quality, plant availability, asset condition and environmental performance.

Many thanks to G. Hijos, J. Artal, C. Pelekani and M. Blaikie for their review of this paper and to the South Australian Water Corporation for their support on the project.