Pretreatment remains one of the most consequential decisions in seawater reverse osmosis (SWRO) design. The way feedwater is conditioned determines not only membrane life but also a plant’s operational stability and cost profile. Sydney Water’s pilot study, presented at the 2007 IDA World Congress, remains a rare example of side-by-side comparison under real coastal conditions, and one that still holds relevance today.
The trial compared ultrafiltration (UF), an emerging membrane technology of its time, with granular media filtration (GMF), a proven dual-media system that combines coagulation and filtration. The goal was simple: determine which could consistently deliver an SDI15 below 3, the benchmark for RO feedwater quality.
How were the systems tested and optimised?
The pilot was fed by open-ocean water from off Sydney’s Kurnell Peninsula: a dynamic environment influenced by the East Australian Current and seasonal phytoplankton blooms. Over nine months, both UF and GMF systems were optimised to achieve their optimal operating conditions.
GMF achieved stability through controlled coagulation at pH 6.5 using 6 mg/L ferric chloride, or a lower dose blended with polyDADMAC. UF, using a ZeeWeed 1000 PVDF membrane, achieved a flux of between 30–42.5 L m⁻² h⁻¹ and required regular chemical cleaning with chlorine and citric acid.
In theory, UF should have produced a more uniform permeate. In practice, the results were more complicated.
What did Sydney Water’s pilot reveal?
Granular media filtration consistently met the RO specification, delivering SDI15 ≤ 3 with low variation. UF achieved similarly low turbidity and near-total microbial removal, yet its SDI fluctuated above 3 for nearly two months.
Investigations ruled out micro-bubble interference and instead traced the cause to feedwater variability. The UF permeate’s SDI closely tracked raw seawater quality, indicating that membrane pretreatment, despite its precision, is not immune to coastal dynamics.
When the UF system was supplied with GMF-filtered feedwater, its SDI dropped from roughly 3.1 to 2.0. The finding was clear: feed stability drives membrane stability.
From a microbiological standpoint, UF excelled. It produced virtually sterile permeate and eliminated nearly all phytoplankton. However, GMF removed about 15 per cent of total organic carbon (TOC) through coagulation: a critical factor in controlling biofouling potential downstream.
This is a subtle but vital distinction. In desalination, achieving low turbidity alone is insufficient. Organic loading, seasonal variation and the interplay between biology and chemistry often determine long-term RO performance more than particulate removal efficiency alone.
What does this mean for future desalination design?
The most surprising outcome was that GMF proved more consistent than UF. While UF offered superior biological removal, its sensitivity to feedwater swings introduced variability that could complicate RO operations. GMF, by contrast, produced steadier SDI results with fewer interventions.
The conclusion, even after nearly two decades, remains relevant. In environments where feedwater conditions fluctuate, such as those along the coast of Australia, robust chemical pretreatment paired with granular media filtration may outperform membranes in terms of reliability and simplicity.
As the original paper summarised, “The GMF plant produced filtered seawater with low and reasonably consistent SDI15, while the UF also had a low average SDI15, albeit with a wider standard deviation.”