Ultrafiltration Advanced Cleaning Research and Modeling Applied to Seawater

By Verónica García Molina, PhD, Guillem Gilabert Oriol and Javier Suárez Martín

Abstract
The well-documented technical advantages of pressurized ultrafiltration (P-UF), compared to alternative conventional pretreatments (such as dual-media filtration), have positioned P-UF as the preferred, most efficient technology in a wide variety of water treatment applications. Continuing advancements in hollow-fiber membrane performance, module design and operating protocols furthers the differentiated cost effectiveness of P-UF. In this article, special focus is given to the optimization of the various processes included in the operation of any UF system in order to attain the maximum efficiency and ultimately, the lowest cost of water. A seawater desalination installation with an ultrafiltration system as pretreatment has been used, first to optimize its efficiency and second, to develop a model to enable the prediction of transmembrane pressure (TMP) increase with time, depending on the cleaning strategy followed. Results of the modeling suggest that the lowest cost of water is achieved through operational schemes based on low frequency of clean-in-place (CIP) and backwash. According to the results obtained by this work, the optimized operation of the UF system can lead to an efficiency increase of the process from 88 to 96 percent. This can be translated into a total cost of water reduction of around four percent.

Introduction
A major difference between RO and UF, apart from the most common construction (spiral wound versus hollow fiber) and the filtration mechanism that dominates each one of the processes (solution-diffusion versus size-exclusion filtration), is the duration of the filtration cycle. More specifically, RO units can be operated continuously for months and are typically stopped for an off-line cleaning, namely CIP, when the normalized permeate flow loss accounts for 10 to 15 percent of the initial flow or a higher than expected differential pressure and/or normalized salt passage is observed. On the other hand, a UF unit typically has filtration cycles between 15 to 90 minutes, depending on the raw-water quality and process design.

In addition to CIP type of cleanings, which are done in both RO and UV technologies in UF processes, backwash (BW) and chemical-enhanced backwash (CEB) are also part of the cleaning strategy. Cleaning more often, or shorter filtration cycles, are required in UF systems compared to RO installations, given the more challenging nature of the water to be treated by UF. Short-term cleanings (often simply called backwash but also backpulse or simultaneous air-scrub reverse flush) are named by the key element in the sequence. They usually occur every 15 to 90 minutes, with a total protocol duration of one to five minutes. Backwash cleanings use an automatic protocol of multiple steps that can vary in terms of order, duration, hydrodynamics and chemistry, typically triggered by time, and sometimes by increases in TMP. Backwash often includes draining, air scouring, backwash (reverse flow) and flushing. Occasionally, small amounts (two to 30 mg/L) of chemicals (especially chlorine) are dosed in the regular backwash. The use of H2O2 (about 25 mg/L) and NaClO (about 10 mg/L) have been studied in the past, concluding that chlorine was far more effective.¹

Medium-term cleanings occur in the range of multiple hours to days (typically four hours to two days) and are applied with an increased concentration of chemicals (especially chlorine, typically 30 to 500 mg/L). Lower concentrations (i.e., in the 50 mg/L range) are used when cleanings take place more frequently (approximately every two to eight hours²), while higher concentrations are applied less frequently, with a range of 12 or more hours. This type of cleaning, which is carried out for an extended time period of 10 to 30 minutes, has been termed chemical-enhanced backwash, maintenance clean or (heated) enhanced flux maintenance. The main differences between a backwash and CEB are the use of higher concentrations of chemicals in the CEB and soaking time, which is a step that takes place only in the CEB to enable sufficient contact time between chemicals and fibers for better cleaning efficiencies. CEB can be classified as acidic or basic, depending on the chemical used. Experience with NaOH (with or without NaOCl) can be found in the literature, but should be used carefully (especially in seawater applications) due to its scaling nature.³ In fact, precipitations have already been discovered with NaOCl, which is also a weak base.⁴ For acidic CEB, typically H₂SO₄ and HCl are used and occasionally, citric acid.

CEB are the most diverse among all cleaning conditions and many different variations have been described. A protocol that combines chemical dose for only a very short time period with air scour has also been proposed.⁵ With outside/in technology, it has also been frequently described to automatically dose chemicals to the feed instead of the product and reuse the solution.⁶ A special backwash protocol involving the use of heated cleaning solution (not only in CIP, but already in CEB) has also been proposed.^7,8 This advanced method has also been described for medium-term cleanings, called heated enhanced flux maintenance (HEFM)—at the Buzzer platform and the Brownsville pilot. This method is used daily. Each microfiltration (MF) rack is taken offline and heated chlorine solution (at about 250 to 400 mg/L chlorine at 30-35 ºC) is automatically circulated through the MF membrane rack for about 30 minutes.^8,9 Some CEB-type, medium-term cleanings may have characteristics of a CIP operation; e.g., involving multiple hours of soak duration and higher concentration.

CIP or long-term cleanings typically occur in the range of several to many months (three to 12) and involve very intense cleaning sequences with high chemical concentrations, high temperature and the use of RO permeate (if available) to prepare the cleaning solution. It is important to emphasize that backwash and CEB typically use UF-filtered water to perform the cleaning. The various cleaning alternatives associated with the UF process offer a wide variety of possibilities for improvement to the overall process in order to optimize recovery, availability, efficiency and chemical consumption. Optimization of an ultrafiltration system can become even more challenging and interesting if all the steps included in each cleaning sequence (air scour, backwash top, backwash bottom, forward flush, CEB, CIP) and all the potential factors to be optimized (flux, chemical concentration, duration and frequency) are taken into consideration. In this article, optimization of the cleaning strategy of a UF system treating seawater is described based on real operating data and on a model that enables the prediction of fouling trends and thus, the need for cleaning. The ultimate goal is certainly to achieve the minimum cost of water through optimized operating sequences.

Materials and methods
As previously stated, the operating conditions of an ultrafiltration system used as a seawater pretreatment for RO are to be optimized in this work. This optimization is based on modeling and simulation of various scenarios, with further evaluation of the operating sequence in the overall sustainability of the process (in terms of fouling increase with time). The operating sequences (mainly cleaning strategies) are tested in two real UF systems, both operated with the same feed seawater. The units used for this experimentation are described below.

Feedwater characterization
The desalination system used for this work, consisting of pressurized, hollow-fiber, outside/in UF and spiral-wound RO, are located and operated in Tarragona, Spain. Feedwater comes from the Tarragona Industrial Harbor in the Mediterranean Sea, where an intake supplies seawater to various industries or companies in the area. The intake has a capacity of 10,000 m³/h but the flow used by the unit is limited to 90 m³/h of capacity. The location of the intake, inside the port of Tarragona, made feedwater quality arriving in the units rather fluctuating. On one side, leakages of oil and petrol from ships and various industrial activities can occur, and on the other, the Francolí River ends in the harbor, which only contains water during heavy rainy days. Sudden increases in TSS content have been monitored in the feedwater from the Francolí River in the days following rain events. Historical values of feedwater TOC and TSS indicated that most of the time, TOC was below five mg/L, while some measurements between five and 10 mg/L have been reported as well. One-hundred percent of the water analyzed during the second half of 2011 and first month of 2012 indicated TOC values below five mg/L. Regarding measurements of suspended solids, the historical data shows an important fluctuation or variability, with the lowest values close to zero and the highest close to 50 mg/L. More recent data showed TSS content below three mg/L (January-February) but peaked up to 20 mg/L during March as a result of the heavy rains that occurred in the region at that time. Regarding turbidity measurements, online turbidity meters were used to monitor feed and filtrate quality, respectively. Feedwater measurements indicated that turbidity is below three NTU 95 percent of the time and below 15 NTU 100 percent of the time.

Installation overview
An installation consisting of UF and RO systems containing two parallel/independent lines has been used for this work. The unit consists of pretreatment (self-cleaning filter) and two identical UF+RO lines (each composed of UF, filtrate tank, cartridge filter and RO) with six four-inch elements in series. The unit also has a CIP system that could be used for either UF or RO (Figure 1 shows a scheme of the installation). Further detailed information regarding the pilot unit itself can be found in existing literature.¹⁰

Figure 1. Installation overview

Results and discussion
Operational data indicated fouling trends corresponding to each set of operating conditions will be shown in terms of TMP increase. The economical evaluation of each operating strategy will be evaluated by directly calculating the cost of water in some cases or by the estimation of those parameters directly impacting the cost of water, namely the chemicals consumption and the availability/recovery/efficiency of the system. The first parameter to be considered is availability, which gives an idea of the time invested in the production of water with respect to overall time (i.e., filtration time + time invested in cleaning). The second important parameter is recovery, which gives an estimation of the water produced versus the feedwater consumed. Efficiency is described as the multiplication of availability and recovery and gives an idea of how efficient the process is, taking into consideration time and production. Finally, in order to be able to make a fair comparison between various operating sequences with different cleaning protocols, the continuous chemical concentration concept is used. This parameter is defined as the amount (in weight units) of chemicals used per unit of volume (feedwater). This parameter could also be used with feedwater flow instead of volume (see Figure 2).

Baseline operating conditions used for this work are 70 L/m2h of operating flux and filtration cycles of 30 minutes. The backwash sequence consists of an initial air scour (30 seconds) followed by draining (10 seconds), then backwash top (130 L/m2h) with air scour (10 seconds), 15 seconds of backwash bottom (130 L/m2h) and finally, a forward flush with a duration of 15 seconds. CEB takes place once per day and a concentration of 350 mg/L of NaClO is used. The sequence of the CEB consists of an initial air scour and backwash as previously described, then a backwash with chemical dosing for 30 seconds, then a soaking period of 15 minutes and finally, another backwash to ensure all the chemicals are removed from the unit before filtration is started again. These operating conditions, considered as a baseline case, are summarized in Table 1. Time consumed by the valves to change position, as well as time consumed by the pumps to achieve the set points are not included in the times mentioned above but they are definitely considered in the calculations, which are shown later.

According to these operating conditions, the recovery of the system was 96.33 percent, the availability 91.1 percent, efficiency 88.46 percent and finally, the equivalent continuous chemical concentration was 0.48 mg/L. This was considered the starting point of this optimization work. More than nine months of UF operational data on treating seawater has been evaluated and used to develop the UF Operation Model, which is aimed at predicting the fouling behavior of an installation or TMP evolution with time. The model was composed of three major segments:

•    Segment A: model of TMP increase with time
•    Segment B: model of TMP decrease as a result of a backwash
•    Segment C: model of TMP decrease as a result of CEB

Equations for each of these segments have been developed empirically in order to be able to predict TMP, with time depending on the operating strategy followed. For example, the model should predict different TMP increase trends depending on the frequency of backwash and/or chemical-enhanced backwash. The methodology consisted of selecting a certain BW and CEB frequency and then letting the model calculate how often a CIP should be applied. The criteria established for this particular exercise was that a CIP should be done every time the model predicted a TMP above 1.7 bar. According to this, the input for the model was: frequency of BW and CEB and maximum allowed TMP. Output was the number of CIPs. See Figure 2 for results of the modeling applied to the various cases evaluated.

The next step in the evaluation was to compare each of the nine operating schemes previously modeled. For this purpose, efficiency (recovery and availability) together with continuous, equivalent chemical concentration was calculated. Total cost of water (including capital and operating expenses) was calculated as well. Capital expenses included feed pumps, automatic strainers, UF skids, CEB/BW and CIP stations, air blowers, electrical installation, control systems instrumentation and civil work. Operational costs included energy and chemical consumption, maintenance cost, labor cost of operating and maintenance and membrane replacement. The cost of water has been calculated for a seawater installation with capacity for 150,000 m3/day. Cost of electricity has been estimated in 0.10 (USD) per kwh and the cost of sodium hypochlorite, is 0.30 per kg. (The results of these calculations are shown in Table 3; efficiency and chemical concentration of each case is plotted in Figure 3.)

Figure 3. Efficiency and chemical equivalent concentration

The first important conclusion from evaluation of the various operating strategies was that as soon as the frequency of the backwash decreased, the efficiency of the process increased significantly. Three different groups of cases have been studied: Cases I and II with BW every 30 minutes; Cases III to VI with BW every two hours and Cases VII to IX without BW. At this point, it should be emphasized that all the cases are based on a sustainable operation, meaning that the TMP will always be maintained below the targeted value. According to the calculations, if BW took place every 30 minutes, efficiency of the process was in the range of 88 percent (baseline scenario), whereas chemical concentration was rather low (below 0.5 mg/L). If the frequency was reduced to once every two hours (Cases III to VI), the efficiency increased up to 96 percent. In these cases, more frequent chemical cleanings were required to maintain a low TMP. Case III was based on frequent CIP and no CEB but this strategy led to relatively high chemical consumption. Cases IV and V were also based on frequent CIP but implemented CEB. The comparison of these two case (IV and V) results were of interest, since by decreasing the number of CIP (from nine to seven in three months) and slightly increasing the number of CEB (from one to two per week), an increase in efficiency from 95.6 to 95.7 percent and a reduction in the chemical concentration from 1.9 to 1.5 mg/L was attained. Following this trend, a further reduction in the number of CIP led to case VI, which has CEB every day but no CIP in 90 days. These conditions led to high efficiencies (96.24 percent) and very low chemical concentrations 0.24 mg/L). If BW wasn’t done (Cases VII to IX), very frequent CEB and/or CIP was required. A strategy based on too frequent CIP might have led to very high chemical consumption, such as in Case VII. On the other hand, a strategy based on frequent CEB led to lower chemical consumption but still quite high efficiency as in Case IX.

According to the previous discussion and to data presented in Table 3 and Figure 3, the two optimum cases were VI and IX. The baseline operating strategy consisted of backwash being done every 30 minutes and one CEB per day. This setup led to an efficiency of 88 percent and an equivalent continuous chemical concentration of 0.48 mg/L. Case VI, based on a backwash taking place every two hours, represented an efficiency of 96 percent and lower chemical consumption, compared to the baseline case. Case IX, which was based on the elimination of backwash, had an efficiency of 97.65 percent but the chemical consumption was comparable to that of the baseline case. The calculation of the theoretical cost of water for each one of the cases modeled indicated that the lowest cost of water was achieved by these two scenarios (Cases VI and IX). A common feature was that frequency of BW and CIP was minimal and sustainable operation was, however, attained through relative higher CEB frequencies.

Conclusions
Operational data collected from nine months of operation of a system treating Mediterranean seawater has been utilized to empirically develop a mathematical model to be able to predict operational trends. More specifically, the model was capable of predicting TMP evolution over time, depending on the process strategy selected. Once the model was validated, it was used to predict TMP with time according to various operating protocols with different frequencies of backwash and chemical-enhanced backwash. When the maximum TMP allowed, selected by the user, was predicted by the model, then a CIP needed to be done. According to this, the output of the model was TMP evolution and the number of CIP to sustain operation. A baseline and nine alternative scenarios or cases have been studied in the framework of this project. The results of the model for each case were used to calculate efficiency and chemical consumption. The cost of the water produced was then calculated according to these parameters. The modeling of each case and the later calculation of the cost of water suggested that the lowest cost in the UF process of seawater was attained when the frequency of the backwash and CIP were reduced to their minimums, while frequency of chemical-enhanced backwash needed to be adjusted accordingly in order to avoid exceeding the maximum allowable transmembrane pressure. The results of this project indicate that this optimized operating sequence resulted in cost of water reductions of five percent and process recovery increases of eight percent.

References

1. Vial, D. and Dousssau, G. The use of microfiltration membranes for seawater pre-treatment prior to reverse osmosis membranes. Desalination 153 (2002) 141-147.
2. Bu-Rashid, K.A. and Czolkoss, W. “Pilot Tests of Multibore UF Membrane at Addur SWRO Desalination Plant, Bahrain. Desalination 203 (2007) 229-242.
3. Glueckstern, P.; Priel, M. and Wilf, M. Field evaluation of capillary UF technology as a pretreatment for large seawater RO systems. Desalination 147 (2002) 55-62.
4. Mourato, D.; Singh, M.; Painchaud, C. and Arviv, R. Immersed membranes for desalination pre-treatment. International Desalination Association (IDA) World Congress, Bahamas, 2003.
5. Vial, D.; Doussau, G. and Galindo, R. Comparison of three pilot studies using Microza membranes for Mediterranean seawater pre-treatment. Desalination 156 (2003) 43-50
6. Rapenne, S.; Port, C.L.; Roddy, S.J. and Croue, J.P. Pre-treatment prior to RO for Seawater desalination: Sydeney pilot-scale Study. International Desalination Association (IDA) World Congress, Maspalomas, Gran Canaria, Spain, October 21-26, 2007.
7. Leal, J.; White, J.M. and Dietrich, J.A. Seawater Desalination in Brownsville, Texas. International Desalination Association (IDA) World Congress – Atlantis, The Palm – Dubai, UAE November 7-12, 2009
8. Boudinar, M.B. ; Choules, P. and Mack, B. Membrane (MF & UF) Pre-treatment Design & Operational experience from three seawater RO Plants. International Desalination Association (IDA) World Congress – Atlantis, The Palm – Dubai, UAE November 7-12, 2009.
9. Brownsville Public Utilities Board, on behalf of Norris, J.W. (NRS), Final Pilot Study Report – Texas seawater desalination demonstration project. Browsville Public Utilities Board, Texas Water Development Board, October 2008.
10. García-Molina, V.; Galvañ, C.; Serallach, X.; Gasia Bruch, E. and Rubio, P. Sistemas Integrados Ultrafiltración-Ósmosis Inversa Aplicados A La Desalación De Agua De Mar. Asociación Española de Desalación y Reutilización (AEDYR) Conference, Barcelona, September 2010.

About the authors
Verónica García Molina, Global Application Development Leader, Dow Water & Process Solutions, holds a PhD in chemical engineering, a Master of Science Degree in chemical engineering and a Master of Science Degree in technological chemistry. She began her career at Dow Water & Process Solutions in 2006 and is currently the Global Application Development Leader.

Guillem Gilabert Oriol, Application Development Engineer, Dow Water & Process Solutions, studied chemical engineering at Universitat Rovira i Virgili (Tarragona) and earned a Master’s Degree in chemical engineering and processes. Currently, he is doing his PhD research and developing applications for ultrafiltration and RO membranes systems for seawater.

Javier Suarez, DOW Ultrafiltration Technical Service Specialist and Market Development for EMEA, Dow Water & Process Solutions, holds a Master of Science Degree in chemical engineering. He has more than 11 years of experience in ultrafiltration and RO water treatment. Suarez has been with Dow since 2007.