Water Journal : Water Journal April 2011
membranes & desalination refereed paper technical features 102 APRIL 2011 water proceed with the known but slightly more expensive MMF and advised the client of this solution. The MMF was installed and commissioned in late July 2006. The SDI returned to <1 upstream of the cartridge filter. Experience with operating the MMF indicated daily backwashing and monthly air scouring were required to ensure mud balls did not form. The long operating period of more than seven months with high iron and cartridge filter as pre-treatment resulted in the leading RO membranes being unrecoverable via CIPs. All of the membranes in one train, Train A, were replaced and the good membranes from this train were used to replace the severely fouled ones in the other two trains. The next challenge, from iodine damage, was unexpected and a first for Osmoflo. After membrane replacement and operation overnight, Train A showed a sudden 35% decline in flux (that flat-lined) and a slight increase in salt rejection. The other two trains did not show a significant flux loss and it was thought that the heavy iron fouling layer reduced the effects of iodine damage. Different CIP regimes using typical chemicals were attempted but were unable to recover Train A's performance. A membrane was removed and some cut-out coupons were sent to Dow Filmtec for autopsy, while other coupons were scanned locally using SEM-EDS (scanning electron microscopy and energy dispersive x-ray spectroscopy) analysis. Both the local SEM-EDS and Dow's surface scans confirmed that iodine was present on the membranes as shown in Figure 5. Dow also tried to wash and scrub off the foulant, but the scan still showed iodine indicating a chemical bond -- i.e., that iodination of the polyamide layer had occurred. It was advised to replace the membranes as they had irreversible flux loss. Dow has only known a small number of cases of iodine damage in the world and this was the first case in Australia that Dow has come across. It is known that the presence of potassium iodide increases the solubility of iodine. It is believed the iodine was a result of the presence of dissolved oxygen and iodide in the groundwater. Osmoflo has now incorporated iodide as a mandatory parameter for all groundwater applications. The basic options for prevention of iodine damage included activated carbon filtration and reduction via sodium metabisulfite (SMBS) dosing upstream of the RO membranes. The caustic (post-treatment) dosing system was converted in September 2006 to an SMBS dosing system that dosed into the RO feedwater as an immediate solution. The SMBS was expected to reduce iodine to iodide, which was well rejected by the membranes. As permeate pH correction was required, a calcite filter was installed later to ensure compliance with treated water quality. The membranes were not immediately replaced, as Osmoflo believed there was a way to restore the flux. Even though actual water usage after mine site optimisation was lower than the warranted capacity, the warranted capacity could be met (with the flux loss) by running all three trains at once instead of just two trains. Fortunately, in December 2006, Dow advised that they have had a few cases where a 5-8% nitric acid soak overnight or a 2% soak for a few days had restored the flux loss. The exact mechanism for nitric acid's effectiveness in this case is not known, but it is postulated that the membrane pores are opened to restore flux, without destroying the polyamide layer and, hence, lower salt rejection. However, multiple CIPs with nitric acid will lead to shorter membrane life. Two membranes were taken out and tested in Osmoflo's workshop. A 5% nitric acid soak overnight was trialled on one membrane and a 2% soak over three days was trialled on the other membrane. Both methods yielded similar flux recovery and so the 5% overnight soak was adopted on the plant as it resulted in less downtime. A 10% solution was also tried, but it rapidly damaged the membranes (membranes turned orange and the anti-telescoping device softened) . The last water chemistry issue Osmoflo experienced at the Ginkgo plant was scaling. In November 2007, the (only duty) antiscalant dosing pump on Train B failed and shortly after the permeate conductivity increased. Due to the pressure to continue to produce water with no immediate available resource for site attendance, the train was allowed to continue with production. Within a few days the feed pressure and permeate conductivity increased dramatically and the train had to stop. Site attendance found white crystals in the RO vessels (Figure 6) and reject pipework. This was later confirmed to be gypsum (calcium sulfate dihydrate) via x-ray diffraction. The severe gypsum scale on the membranes could not be removed via CIP and most of the membranes had been mechanically damaged by the crystals, so all of them had to be replaced on the train. Following this experience each train was upgraded to have standby pumps and flow monitors (switch) for antiscalant and SMBS dosing, which are critical to operations. Overcoming the challenges with water quality and remote operations in the first two years was beneficial for Osmoflo. The filter cartridge replacement and RO membrane CIP have returned to expected frequencies of monthly and quarterly, respectively. There have been rare cases of SMBS dosing failures (at most, once a year), but the nitric acid CIPs have always restored flux performance. Membranes have been replaced on an as-needs basis when salt rejection has declined to an unacceptable level. The average membrane life has been estimated to be greater than three years. The BOO contract was extended by another 18 months by Bemax in 2010. Figure 5: SEM-EDS of a damaged membrane. Figure 6: Tail end view of gypsum scaling.
Water Journal March 2011
Water Journal May 2011