EVALUATION OF MF/UF CHEMICAL CLEANING STRATEGIES IN DIRECT POTABLE REUSE APPLICATIONS Chelsea M. Francis, Arcadis, 401 E Main Dr Suite 400 El Paso, TX, 79901 E-mail: chelsea.francis@arcadis.com Phone: (602) 403-9287 Brent Alspach, Arcadis, Carlsbad, CA John Balliew, El Paso Water Utilities, El Paso, TX Introduction With an already diverse portfolio of water resources becoming increasingly strained by a prolonged drought, El Paso Water Utilities (EPWU) is developing a direct potable reuse (DPR) project to recycle 10 MGD of secondary effluent from its Roberto R. Bustamante Wastewater Treatment Plant (WWTP). Upon completion, the new Advanced Water Purification Facility (AWPF) will be only the second reuse treatment plant in the world (after the Goreangab Water Reclamation Plant in Windhoek, Namibia) to pump highly purified recycled wastewater directly to the potable water distribution system. While interest in both direct and indirect potable reuse is rapidly expanding in the US and other locations around the globe, there are very few such full scale systems in operation worldwide. As a result, institutional knowledge about potable reuse remains limited relative to other water and wastewater treatment applications. For example, EPWU is currently conducting a pilot study on the treatment train proposed for the full scale AWPF, which includes membrane filtration. Systems from two leading MF/UF suppliers are being tested in parallel, and while both suppliers have similar experience in potable reuse applications in Texas (pilot and/or full scale), and both utilize encased PVDF membranes, the proposed chemical cleaning strategies are significantly different. One supplier begins by applying caustic and sodium hypochlorite and then finishes with a citric acid step; the other supplier reverses this order of chemical application, starting with an acid cleaning step (hydrochloric + citric) and then finishes using sodium hypochlorite. Notably, the latter supplier does not utilize caustic, which is commonly applied to remove organic fouling, even though total organic carbon (TOC) levels in the MF/UF feed are routinely on the order of 10-15 mg/l. Consequently, the EPWU DPR pilot affords a unique opportunity to evaluate different approaches to MF/UF cleaning in potable reuse applications, thereby advancing the institutional industry knowledge. At least six clean-in-place (CIP) events will occur over the nine-month duration of the pilot program, and the efficacy of both suppliers strategies is being tracked using the clean water flux (CWF) test as described in the United State Environmental Protection Agency (USEPA) Membrane Filtration Guidance Manual. The collected results of these tests will indicate: the effectiveness of each respective cleaning step (and thus which foulants are most prevalent organic, inorganic, and/or biological); the overall effectiveness of each CIP event for restoring baseline permeability; the accumulation of chemically irreversible fouling; and the comparative success of the two suppliers CIP strategies. This paper and associated presentation evaluate and discuss the results of this testing, commenting on optimal CIP strategies for potable reuse and augmenting the industry knowledge base. 1
AWPF Background and Summary In 2012, EPWU initiated a feasibility study to assess the potential for indirect potable reuse (IPR) near its Bustamante WWTP, the Jonathan Rogers WTP, and the Rio Bosque Wetlands Park (adjacent to the plants). The IPR concept was to divert a portion of the effluent from the Bustamante WWTP and treat it for use as an additional supply to the Jonathan Rogers WTP, augmenting available supply from the Rio Grande River. However, EPWU discovered significant challenges due to local hydrogeological limitations and recognized the advantages of instead pursuing DPR. A significant advantage favoring a DPR concept included leveraging the unusual proximity of the wastewater treatment plant and the existing water distribution infrastructure at the site. In addition, advanced treatment and process monitoring technologies were readily available enabling progress in DPR applications in the water industry. Based on feasibility assessment results, EPWU proceeded with the process of pilot testing and subsequent implementation of a 10 MGD (influent) AWPF to realize the advantages of DPR in its efforts to enhance water supply reliability. The source water for the AWPF (both pilot and full scale) is unchlorinated secondary clarifier effluent from Bustamante WWTP. The Bustamante WWTP is an activated sludge plant that treats an average dry weather wastewater flow of approximately 29 MGD, comprised of both domestic and industrial wastewater discharges, with the latter accounting for approximately 11% of the total flow. Table 1 provides a summary of the Bustamante WWTP effluent historical water quality from a period of January 2011 to May 2014. Table 1. Bustamante WWTP Effluent Historical Water Quality (January 2011 to May 2014) Parameter Min 95th % Max ph 6.6 7.0 7.1 Alkalinity, mg/l as CaCO3 29.0 214 244 Turbidity, NTU 1.0 8.2 30 BOD5, mg/l 2.0 9.5 14.8 Ammonia, mg/l as N 0.3 24.4 35.0 Nitrate, mg/l as N 0.5 22.7 33.0 Nitrite, mg/l as N 0.07 3.9 5.9 Total Dissolved Solids (TDS), mg/l 566 1210 1250 Sodium, mg/l 147 325 366 Chloride, mg/l 114 382 538 Calcium, mg/l 31.2 77.8 86.7 Magnesium, mg/l 8.2 18.8 36.9 Sulfate, mg/l 97.4 307 543 Orthophosphate, mg/l 0.4 4.2 7.0 The AWPF pilot plant processes include membrane filtration (microfiltration (MF) and ultrafiltration (UF)), membrane desalination (nanofiltration (NF) and reverse osmosis (RO)), ultraviolet disinfection with advanced oxidation (UV-AOP), and peroxide-quenching granular activated carbon (GAC). In addition to the treatment train, multiple online analyzers were 2
strategically placed to evaluate the monitoring techniques essential to public health risk reduction. These instruments include analyzers for TOC, nitrite, nitrate, ammonia, UV-254, turbidity, free chlorine, total chlorine, conductivity, ph, and ultraviolet light transmittance (UVT). A full programmable logic controller (PLC) was employed to give the pilot plant full automation equipped with a SCADA system to collect data from online analyzers and treatment processes. The AWPF Pilot Plant allows for the analyzers, controllers, and programming to be tested prior to implementation for the full-scale facility. Secondary clarifier effluent is sent to a break tank and pumped through ozonation treatment during a portion of the pilot test and directly through the MF/UF membrane systems during other periods of operation. Pretreatment for the MF/UF membranes consists of chloramines and strainers to minimize biological and particulate fouling, respectively. The target chloramine dose will be evaluated during initiation of pilot testing and is expected to range from 2 to 4 mg/l. The two MF/UF systems being tested will be provided by Evoqua Water Technologies and Pall Corporation. Table 2 provides characteristics for the MF/UF membranes. Table 2. MF/UF Module/System Characteristics Criteria Evoqua Pall Module model L40N/CPII UNA-620A Flow pattern Outside-In Outside-In Recovery (%) 92-96 95-98 Nominal pore size (µm) 0.04 0.1 Membrane material PVDF1 PVDF1 Membrane Surface Area (ft2) 721 538 Number of Modules 2 4 Total Surface Area (ft2) 1442 2152 Anticipated Flux (gfd) 25 25 Pilot System Flow (gpm) 26 37 1 PVDF - Polyvinylidene fluoride Key Messages The data analyzed in this project prove the CWF test as a valuable diagnostic tool in determining CIP effectiveness and whether irreversible fouling has occurred. With a limited source of data on this subject this paper proves useful for industry professionals running or tracking MF/UF systems such as utilities/system owners and operators, regulators, design engineers, and manufacturers. The AWPF results indicate the effective cleaning step (and thus most prevalent foulants organic, inorganic, and/or biological); the effectiveness of each CIP event for restoring baseline permeability; the accumulation of chemically irreversible fouling; and the comparative success of the two suppliers CIP strategies in water reuse applications. A CWF test should be conducted at start up and during every CIP to track performance from baseline conditions. This is an easy test that can be completed quickly during CIP process. 3
Normalizing temperatures prevents water viscosity changes from increasing the measured TMP at various temperatures (USEPA Membrane Filtration Guidance Manual (Sec. 2.4.2)), which would otherwise show an artificially high indication of fouling. CWF tests performed between CIP steps are an indicator of what foulants are present and which cleans are more effective. Acid / low ph removes inorganic scaling and particulates, while a basic / high ph step removes organic fouling and chlorine step removes biological fouling. Steeper sloped CWF trends may indicate lower permeability, and a progressive trend may suggest irreversible fouling. MF/UF Suppliers Clean-in-Place Strategies The AWPF has the distinct advantage of testing not only two MF/UF membranes but two different cleaning strategies with the same influent water quality in a potable reuse application. CIPs are performed to remove any fouling on the membranes and the CWF test is used to trace reversible and irreversible fouling by recording the transmembrane pressure (TMP) over various fluxes or flows of filtered water at a normalized temperature for the specific application. TMP is affected by the viscosity of the water with an increased resistance to flow at cooler temperatures; thus, when comparing the results from multiple CIPs, temperature correction is important to avoid viewing the increased resistance as fouling rather than a viscosity increase. The TMP is plotted versus the flux or flow with the slopes indicating the permeability of membranes (i.e., steep slopes result from more resistance to flow with a higher degree of fouling (low permeability), and small slopes from less resistance to flow with a lower degree of fouling (high permeability). The data presented in this manuscript are normalized to 20 C. The CWF tests can inform relative CIP effectiveness at each step in the process, as well as overall cleaning efficacy compared to the baseline. This allows for adjustments to be made in the cleaning strategy and gives an indication of the type of foulant(s) present. Pall Corporation performs CIPs by beginning with a chlorine/caustic blend to remove any biological/organic fouling then performs a second step with acid to remove any inorganic scaling or particulates. Evoqua sequences the CIP by beginning with an acid step to remove any inorganic scaling or particulates and follows that up with a chlorine step to remove organic and biological fouling. Some differences of note relative to the Pall approach are that Evoqua uses influent water, not filtered water, for the CWF test, and the high ph clean utilizes only chlorine and no caustic. The use of influent water to perform the CWF tests results in a higher TMP, and increases the slope of the plotted results. Since the CWF test compares performance to a baseline set of results the test remains a means of tracking membrane fouling. 4
CWF Results Figures 1-7 show all the CWF results for the Pall system from July 13, 2015 to the latest CIP November 11, 2015. Figure 1 represents the Pall baseline CWF results with a post CIP slope of 0.054 psi/gpm. The CIPs on July 29 th and August 26 th (Figures 2 and 3) were conducted after major fouling events and in both cases the CIPs were effective and returned the slope to baseline conditions. With the exception of the two CIPs mentioned above, all others were performed before terminal temperature corrected TMP levels were reached and the fouling was minimal. The following figures show a majority of the decrease in slope after the chlorine and caustic step with little recovery following the acid step. Indicating the chlorine/caustic step is most effective and the biological and organic fouling accounting for the majority of reduction in permeability before the CIP. Figures 8-14 show all the CWF results for the Evoqua system from July 16, 2015 to the latest CIP November 11, 2015. Figure 8 represents the Evoqua baseline CWF results with a post CIP slope of 0.105 psi/gpm. The fouling events in July and August were followed by a modified CIP, once the system reached terminal TMPs the membranes were soaked in chlorinated water until the CIP was performed (results shown in Figures 9 and 10). Starting with the acid step is effective in returning half the permeability and the chlorine step returns the membranes back to baseline conditions or better. The chlorine and acid steps are equally effective with an average of 0.05 psi/gpm recovery each step, suggesting biological and/or inorganic fouling. The recent CIPs permeability did not have a noticeable change since the slope of the CWF test after the CIP is similar to the slope before the CIP. The two different cleaning strategies portray two different fouling potentials. Pall cleaning with chlorine/caustic step then acid step indicates there is biological and organic fouling. Evoqua starting with an acid step and finishing with chlorine alone indicates both biological and inorganic fouling (no caustic used to indicate organic fouling). 5
Figure 1. Pall Baseline CWF Test Results (July 13, 2015) 6
Figure 2. Pall CWF Test Results (July 29, 2015) 7
Figure 3. Pall CWF Test Results (August 26, 2015) 8
Figure 4. Pall CWF Test Results (September 8, 2015) 9
Figure 5. Pall CWF Test Results (October 12, 2015) 10
Figure 6. Pall CWF Test Results (October 26, 2015) 11
Figure 7. Pall CWF Test Results (November 11, 2015) 12
Figure 8. Evoqua Baseline CWF Test Results (July 16, 2015) 13
Figure 9. Evoqua CWF Test Results (July 28, 2015) 14
Figure 10. Evoqua CWF Test Results (August 25, 2015) 15
Figure 11. Evoqua CWF Test Results (September 8, 2015) 16
Figure 12. Evoqua CWF Test Results (October 12, 2015) 17
Figure 13. Evoqua CWF Test Results (October 26, 2015) 18
Figure 14. Evoqua CWF Test Results (November 11, 2015) 19
Evidence of Irreversible Fouling? If the CWF test results are tracked from membrane start-up, the baseline performance is compared to the post CIP performance and used to track irreversible fouling. The CWF results do not indicate only irreversible fouling, water quality data and operational changes should be evaluated when CIPs do not return the permeability to the baseline conditions. If the data collected do not explain the reduction in permeability from the baseline then perhaps a more effective cleaner (i.e., stronger dose, different configuration or cleaning solution) would bring the permeability back to baseline conditions. If baseline conditions are not met through these troubleshooting steps then irreversible fouling may have occurred. The Pall and Evoqua CIPs were scheduled to be within a couple days of each other to remain consistent for comparison between cleaning strategy (i.e., exposure to equal fouling potential). To date, seven CIPs have been performed. Figure 15 shows the CWF results after each Pall CIP over the duration of the pilot. The trends for each CIP are overlapping, indicating that no irreversible fouling has occurred with a post CIP range of 0.05 to 0.07 psi/gpm. The CIPs are remaining effective for the operating parameters and water quality tested. After the terminal fouling events subsequent CIPs were unable to return the slope to 0.054 psi/gpm, though they did achieve as low as 0.062 psi/gpm which is very close to baseline. This is a trend to be monitored further before irreversible fouling can be determined. Figure 16 shows the CWF results after each Evoqua CIP over the duration of the pilot. The post CIP CWF results do not seem to follow a trend, the baseline results have a lower permeability than the two CIPs in July and August. The use of influent water (not filtered water) with varying water quality may be the reason for varying results. Also the soak for multiple days in chlorinated water would have ensured a thorough clean for the two CIPs after terminal TMPs were reached in July and August. The post CIP results from September through November are experiencing less permeability recovery, with a total range of 0.087 to 0.132 psi/gpm post CIP results. It is possible the use of caustic may account for the discrepancy between recoveries and may be added to the strategy. The reduction in recovery may be noise from the varying influent water quality so the CWF results will be monitored for continual increases in slope. 20
Figure 15. Pall CWF Results Post-CIP Comparison 21
Figure 16. Evoqua CWF Results Post-CIP Comparison 22
Conclusions The AWPF piloting of the two cleaning strategies provides us with valuable information on the foulants in the secondary clarifier effluent and the effective strategies for managing the membrane operations. It has been proven that the order of cleaning does not reduce the CIP effectiveness though optimization of the step order, cleaning dosages, and chemicals is recommended. The sodium hydroxide/caustic solution shows the most recovery of permeability back to baseline conditions for the secondary clarifier effluent for potable reuse. The foulant of concern in the secondary clarifier effluent is biological and organic material and the correct pretreatment and CIP strategies are capable of managing the membrane fouling. Irreversible fouling did not occur on the membranes and the trends of post CIP CWF results are monitored as they may be reducing permeability. References United States Environmental Protection Agency (2006), Membrane Filtration Guidance Manual, EPA 815-R-06-009. 23