Reclaimed Water Desalination Technologies: A Full-Scale Performance and Cost Comparison
Between Electrodialysis Reversal and Microfiltration/Reverse Osmosis
R. Shane Trussell, Ph.D., P.E., BCEE Principal Investigator Gordon J. Williams, Ph.D., P.E. Project Engineer
Webcast August 9, 2012
1
Webcast Overview
• Background • Previous Studies & Project Motivation
• Facility Comparison • Cost Comparison • Conclusions
2
Background
3
Why Desalt Recycled Water?
• Reduce total dissolved solids (TDS) to expand possible uses
• High salt rejection not necessarily needed – Opportunity to blend membrane product water
• Membrane desalination – Reverse osmosis (RO) with microfiltration (MF)
pretreatment – Electrodialysis reversal (EDR)
4
Cathode (-)
Anode (+)
Product Water
Concentrate
Product Water
Concentrate
Concentrate
RO EDR Pressure
Saline Fresh
Two Desalting Membrane Types
RO has significantly higher salt removal 2-Stage System: RO 90-98% vs. EDR 50-70% 5
RO Membranes
6
Image from: Baker, R. Membrane Technology and Applications. John Wiley & Sons: West Sussex, England (2004)
Types of RO membranes • Cellulose acetate • Thin film composite
RO Vessel Architecture
7
Image from: Baker, R. Membrane Technology and Applications. John Wiley & Sons: West Sussex, England (2004)
EDR Process
WateReuse Research Foundation 11
Figure 1.4 Conceptual schematic of how charged ions are removed from the demineralization flow by passing through anion and cation membranes. When the polarity of the electrodes changes (Polarity Mode 1 compared to Polarity Mode 2), the role of the central channels (concentrate or demineralized) switches.
Polarity Mode 1
Polarity Mode 2 (charge reversed from above) 8
EDR Process
WateReuse Research Foundation 11
Figure 1.4 Conceptual schematic of how charged ions are removed from the demineralization flow by passing through anion and cation membranes. When the polarity of the electrodes changes (Polarity Mode 1 compared to Polarity Mode 2), the role of the central channels (concentrate or demineralized) switches.
Polarity Mode 1
Polarity Mode 2 (charge reversed from above)
9
EDR Membranes and Spacers
12 WateReuse Research Foundation
Figure 1.5 Schematic of EDR spacers (aerial view). A spacer is placed in between each membrane to separate the membranes from each other, providing a flow path for water to pass between the membranes. The four holes at the bottom of the spacer (numbered 1 through 4) go through each spacer and membrane. When the membranes and spacers are stacked together, the holes line up and become conduits for (1) concentrate influent, (2) feed water influent, (3) concentrate effluent, and (4) product water effluent (flow through the holes is perpendicular to the membrane). For each spacer, only two of the four holes connect to the flow path (i.e., holes 2 and 4 on the left and holes 1 and 3 on the right; flow path through the spacer marked with dashed line). The spacer orientation is alternated between configurations A and B, such that every other spacer compartment is the flow path for either the concentrate or product flow. When the polarity is switched, the flow designation of holes 1&2 and 3&4 are switched, so what was once the concentrate flow path, now becomes the demineralized flow path, and vice versa.
Spacers
10
Two types of membranes: • Anion Transfer Membrane • Cation Transfer Membrane
Typical numbers per stack: • 1200 membranes • 1201 spaces
EDR Architecture
WateReuse Research Foundation 13
Figure 1.6 Side view schematic of an EDR stack (arrows indicate flow direction). When the electrode polarity changes, the following designations are switched (1) “feed water” and “concentrate in,” (2) “product water channel” and “concentrate channel,” and (3) “product water” and “concentrate out” (adapted from Meller, 1984).
1.3.2 EDR History and Types of Membranes Although ED has only been implemented in industrial applications for approximately 40 years, the principles of ED were developed more than 90 years ago. In fact, the multi-compartment ED cell with alternating cation and anion membranes that is used in modern applications has its roots in a design published by Meyer and Straus in 1940. At that time, however, the success of ED was limited by a lack of appropriate materials. The development of proper ion-exchange membranes during the 1950s facilitated great improvements in ED feasibility and application (Langelier, 1952). Ionics Inc. developed the first commercial ED process and began selling the process in 1954, with the first installation in Saudi Arabia (Katz, 1977). The company went on to further innovate the ED process by introducing a polarity reversal step, later termed electrodialysis reversal, in order to control scaling (Katz, 1979). The EDR process essentially replaced conventional ED and still retains a significant percentage of the desalination market today. In the 1990s, EDR was successfully applied to demineralizing nonpotable reclaimed water (Reahl, 2005). ED membranes are typically flat sheets of plastic mounted on a fabric backing to provide structural support (AwwaRF et al., 1996). Previous generations of anionic membranes were
11
Comparison Between Membranes Consideration RO EDR
Use in U.S. Recycled Water • Widely applied • 1 Facility (San Diego)
Pretreatment for Reuse • MF required • MF not required
Fouling Concerns
• Inorganic • Organic • Biofouling • Silica
• Inorganic • Organic
O&M
• MF CIP • MF maintenance cleans • RO CIP • Conductivity profiling • Interconnector repair
• EDR CIP • Cartridge filter replacement • Manual membrane cleans • Probing for “hot spots” • Torque adjustment
12
Assuming Two-Stage Systems
13
RO: Two-stages
EDR: Two stages
Influent
Influent
Effluent
Effluent
RO: Two passes Effluent Influent
Previous Studies and Project Motivation
14
Limitations of Previous Studies
• Studies based on modeling, piloting or greenfield startup – None examined longer-term operation
• Costs will differ between brackish water and reclaimed water
• Most missing some of life cycle costs – e.g., membrane replacement, longer-term
maintenance activities
15
16
Previous Studies: Costs per Mass of Salt Removed
*See WRF 08-17 report for full references
Cost per Mass Salt Removed ($/ton of TDS)
O&M Total
Study Water MF/RO EDR MF/RO EDR Port Hueneme Demo
(Leitz, 2001) Brackish $120 $124 $276 $319
San Diego Startup (MWH, 2002) Recycled $56 $66 -- --
San Jose Pilot (Adham et al., 2004) Recycled $213 $218 $494 $303
San Diego Pasqual (MWH, 2008) Brackish $275 $248 $430 $437
Overview of This Study • Motivation: Most existing systems are
RO, but EDR may provide cost savings • Objective: Document actual full-scale
costs of EDR and MF/RO – Equipment and capital, energy, membrane
replacement, labor, and chemicals • Approach: Compare two similar facilities
– Size, established operations, feedwater, location
17
Facility Comparison
18
Comparison of Study Facilities
Parameter MF/RO EDR Location (California) Long Beach San Diego Online Date 2003 1998 Membrane Design Capacity (mgd) 3.0 3.3 Upstream secondary process Nitrified activated sludge Upstream tertiary process Granular Media Filtration Feed Turbidity (NTU) 0.5 0.7 Daily Production (mgd) 2.7 1.5
MF/RO: Leo J. Vander Lans AWT Facility (LVL) EDR: North City Water Reclamation Plant (NCWRP)
19
LVL MF/RO Process Train
Tertiary influent
MF RO (1st Stage)
Backwash Flow Brine
Antiscalant RO (2nd Stage)
Effluent to UV and
Barrier
20
LVL Microfiltration (MF) • Pall Corp Microza • Module: USV-6203 • Hollow-fiber • Nominal pore size: 0.1 µm • 200 modules • 4.2 mgd capacity • 92% design recovery
21
LVL Reverse Osmosis (RO)
• Hydranautics ESPA-2 • Spiral wound • Two stage system
• 72:36 pressure vessels • 3.0 mgd capacity • Overall RO recovery: 85%
22
MF Performance
23 Data represent 3 years of daily grab samples
RO Performance
24
WateReuse Research Foundation 27
Figure 2.10 Typical RO performance at LVL (data from April/June 2009).
2.2.4 RO System Salt Removal The average salt removal from the RO system from March 2003 through July 2009 was 87%, which is uncharacteristically low for an RO system (see Figure 2.11). The low average TDS rejection was a result of a prolonged period of operation with underperforming second stage membranes (January 2007 through early April 2009), where the average salt rejection during this period was 80%. The second stage was underperforming because the membranes were subject to frequent CIPs because of aluminum carryover from the LBWRP and improperly positioned valves that caused backpressure on the second stage elements. Unlike an MF CIP, an RO CIP is completely manual and requires two operators to properly position valves while cleaning each stage. The manual CIP nature of RO systems make these cleans more operator intensive than MF CIPs, and there is also greater potential for human error. The entire second stage was replaced in mid-April 2009, and the salt removal has since ranged from 94–96% of the dissolved solids. Excluding the period of uncharacteristic salt rejection (i.e., averaging data from March 2003 through 2006 and mid-April 2009 through July 2009) the average RO membrane salt rejection was 94%, which is the salt removal value used for comparison with EDR.
The RO membranes in the first stage (504 elements) have been in service since the plant commissioning was completed in 2003 and recent conductivity profiles (March 2008) show the increased salt passage that has occurred over the 5 years (see Figure 2.12A). The first stage product water conductivity has increased from an average of 15 to 40 PS/cm, which is a reasonable increase in salt passage, but is still reflective of the frequent CIPs experienced early on at LVL.
LVL RO performance from April to June 2009
RO Performance
25
MF/RO Recovery
WateReuse Research Foundation 29
Figure 2.12 Conductivity profile probability plot from (A) first RO stage and (B) second RO stage.
2.2.5 Overall MF/RO System Feed Water Recovery
Because MF pretreatment is a requirement for a successful application of RO membranes for wastewater applications, the cumulative losses through both systems need to be considered. The variability in overall feed water recovery at LVL is presented in Figure 2.13. The feed water recovery is typically 80%, which is a combination of an operational recovery of 94% through the MF process and 85% through the RO system.
Figure 2.13 Variability in overall LVL water recovery (combined recovery of MF and RO processes).
26
NCWRP EDR Process Train
27
NCWRP EDR Membranes
• GE/Ionics EDR 2020 • Two stage system • 15 EDR lines
• 30 EDR stacks • 1,200 membranes per
stack • 36,000 membranes
total • 3.3 mgd capacity
28
Pretreatment Performance: GMF
29 Turbidity data from 2005 through April 2009
EDR Performance
30
WateReuse Research Foundation 43
Although operating at a target of 65% salt removal, the actual salt removed was 60% on average and was below the absolute minimum of 55% about 4% of the time (Figure 3.6). The salt removal decreases as the membranes foul, but also varies with changes in the power applied to the stacks. On a time scale of minutes, a 3% decline in salt removal is typically seen as a result of fouling during the 15 min before the charge is reversed (Figure 3.7; salt removal drops from 61% to 58%), but the salt removal consistently rebounds after the polarity switches. On a time scale of weeks, the peak level of salt removal seen after charge reversal will gradually decline. In order to maintain adequate salt removal, operators occasionally increase the power applied to an EDR unit by increasing the voltage. After necessary voltage exceeds 400V, the EDR unit will be shut down and cleaned by salt CIP to restore performance. After cleaning, the voltage requirements are much lower, but eventually, the restorative power of the CIP will diminish and a manual clean is necessary to restore the EDR unit’s performance.
Figure 3.7 EDR current, power, and salt removal over a 6-h period (4/24/09). Total power and current represent all five lines and both stages of a single EDR unit. The electrode charges are reversed every 15 min to prevent scaling on the membrane. Average feed water conductivity was 1770 PS/cm and last CIP was 49 days prior.
Day-to-day, the electrical potential applied to the EDR stacks remains fairly constant. As the stack begins to foul over a matter of weeks, the plant operator will make discrete increases in the voltage to maintain salt removal. During February through March 2009, the electrical potential in EDR Unit 1 was initially 360V, but after 3 weeks of operation, the voltage was increased to 390V. Then after another 2 weeks, voltage was increased again to 415V, requiring a CIP with salt and acid only 3 days later (Figures 3.8 and 3.9). After the CIP, the required voltage to achieve the target salt removal was restored to 360V. These incremental
EDR operation on April 24, 2009
EDR Performance
31
WateReuse Research Foundation 45
Figure 3.8 EDR performance data after manual clean (performed January 30, 2009) through the first CIP event (Unit 1).
Figure 3.9 EDR performance directly before and after a CIP clean (data subset of Figure 3.8).
EDR Performance
42 WateReuse Research Foundation
Figure 3.5 Variability in EDR feed water and product water conductivities.
3.3.2 Salt Removal, Electrical Potential, and Current Unlike RO membranes, salt removal through the EDR process can be controlled to a certain
degree by varying the electrical current. When the NCWRP first began operating the EDR
facility, the system was operated to achieve a salt removal of 70–72%, but this relatively high
salt removal for a two-stage EDR system led to difficulties maintaining the system, such as
more frequent cleanings, and operating problems, such as hot spots. After a few years of
operating at a high salt removal, NCWRP decided to lower the salt removal target by
lowering the electrical potential. They settled on an operating target of 65% with an absolute
minimum of 55%, below which a CIP would be performed to restore the performance. In the
case of the NCWRP, lowering the salt reduction also meant that they needed to construct
additional EDR capacity to meet their salinity reduction goals.
Figure 3.6 Variability in EDR salt removal and water recovery (Unit 1).
32
Membrane Performance Summary
Parameter Units MF/RO EDR Water recovery % 80% 85% Salt Removal % 94% 60% Avg. feedwater TDS mg/L 654 1115 Avg. product water TDS mg/L 39 444
33
Cost Comparison
34
Costs Included in Comparison
• Capital • Membrane replacement • Energy • Labor (maintenance) • Chemicals for operation • Chemicals for cleaning
35 Note: All costs presented are in 2009 dollars
Costs Excluded from Comparison • Footprint costs
– Land acquisitions, site improvements, structure • General construction costs
– Management, engineering & design • Concentrate disposal • Facility operations
– Labor beyond maintenance/cleaning
36
Cost Comparison Assumptions • Everything but membrane type is the same
– Influent TDS, production, salt reduction goals • Operated at design capacity, year round • 20 year facility life span; no salvage value • Electricity: $0.105 per kWh • Labor: $65/h • Present worth calculations
– 3% inflation, 6% discount
37
Costs (normalized by flow)
Cost MF/RO EDR Initial equipment and construction $249 $114 Membrane replacement $20 $26 Energy costs $62 $51 Chemical costs (process) $22 $1.3 Chemical costs (CIP) $3.0 $3.2 Labor cost (maintenance only) $16 $32 Total $371 $228
Dollars per acre-foot of membrane product
38
Costs (normalized by flow)
Cost MF/RO EDR Initial equipment and construction $249 $114 Membrane replacement $20 $26 Energy costs $62 $51 Chemical costs (process) $22 $1.3 Chemical costs (CIP) $3.0 $3.2 Labor cost (maintenance only) $16 $32 Total $371 $228
Dollars per acre-foot of membrane product
39
Salt removals not equal
Normalize costs based on salt removed
Normalized by Salt Removal: A Blending Example
• Assume a system is needed to produce 10 mgd recycled water
• Current TDS is 1000 mg/L, but users need 800 mg/L
40
Normalized by Salt Removal: A Blending Example
• Assume a system is needed to produce 10 mgd recycled water
• Current TDS is 1000 mg/L, but users need 800 mg/L
41
20% salt reduction
Normalized by Salt Removal: A Blending Example
• Assume a system is needed to produce 10 mgd recycled water
• Current TDS is 1000 mg/L, but users need 800 mg/L
42
20% salt reduction
What size membrane system is needed?
Membrane System Size Related to Salt Removal Efficiency
43
Membrane capcity (mgd) = overall salt removal from blendmembrane salt removal
× blended flow (mgd)
Parameter Units MF/RO EDR Salt Removal % 94 60
Membrane System Size Related to Salt Removal Efficiency
44
Membrane capcity (mgd) = overall salt removal from blendmembrane salt removal
× blended flow (mgd)
MF/RO capcity (mgd) = 20%94%
×10 mgd = 2.1 mgd
Membrane System Size Related to Salt Removal Efficiency
45
Membrane capcity (mgd) = overall salt removal from blendmembrane salt removal
× blended flow (mgd)
MF/RO capcity (mgd) = 20%94%
×10 mgd = 2.1 mgd
EDR capcity (mgd) = 20%60%
×10 mgd = 3.3 mgd
Membrane System Size Related to Salt Removal Efficiency
46
Membrane capcity (mgd) = overall salt removal from blendmembrane salt removal
× blended flow (mgd)
MF/RO capcity (mgd) = 20%94%
×10 mgd = 2.1 mgd
EDR capcity (mgd) = 20%60%
×10 mgd = 3.3 mgd
EDR is ~60% larger than the MF/RO
Costs (normalized for 20% salt reduction) MF/RO ($79/ac-ft*) EDR ($76/ac-ft*)
Capital $53
Capital $38
Energy $17
Energy $13
Maintenance $10.5
*In dollars per acre-foot of blended water
Maintenance $3.3
Replace membranes $8.6
Replace membranes $4.2 Chemicals - $5.3 Chemicals - $1.5
Other $13
47
Footprint and Brine Flow from Example • Membrane system footprint:
• Waste flow produced:
48
Additional Factors to Consider • Need to Remove Other Constituents • Presence of Certain Foulants (silica) • Occupational Hazards • Usage: Seasonal vs. Year-Round • Footprint Constraints • Brine Disposal Method
49
A Need to Remove Other Constituents When removal of other constituents is needed
• Comparison based on salt removal only
• RO provides an effective barrier to pathogens and TOC
• EDR does not provide this type of barrier
• MF/RO is well-suited for groundwater recharge
Effectively Removed by MF/RO
TOC particles
pathogens uncharged molecules
50
• Silica fouling a problem for RO
• Silica does not concentrate in an EDR
• EDR is well-suited for high silica applications
51
Presence of Certain Foulants: When high levels of silica are present
Image from: Gabelich, C. J..; Chen, W. R.; Yun, T. I.; Coffey, B. M.; Suffet, I. H. (2005). The role of dissolved aluminum in silica chemistry for membrane processes. Desalination 2005, 180, 307-319.
• Voltage: EDR membranes stack often exposed (e.g., checking for “hot spots”) and improper handling could result in shock – Measures can be taken to minimize risk
• Pressure: both systems under high pressure – RO (~200 psi) and EDR (~50 psi) – Risk low due to overdesign for safety
52
Occupational Hazards: High Voltages and High Pressures
• Many reuse applications are seasonal – e.g., agriculture, landscape irrigation
• Seasonal flow favors EDR, with the lower capital cost and higher O&M – Chemical and energy costs reduced
proportional to flow – Labor and membrane replacements costs
reduced but not directly proportional
53
Usage: Seasonal vs. Year-Round: When the system is only used seasonally
• Cost of acquiring and developing land not considered in comparison: – Footprints similar on flow normalized basis – MF/RO will be smaller for same salt removal
54
Footprint Constraints: When the size of footprint is important
Assuming 10 mgd blended flow with 20% TDS reduction
Impact of Footprint on Blend Cost Comparison (10 mgd with 20% salt reduction)
55
Footprint Constraints: When the size of footprint is important
Footprint Cost ($/ft2)
MF-RO ($/ac-ft blend)
EDR ($/ac-ft blend)
MF-RO:EDR
$0 $78.9 $75.7 1.04 $10 $79.3 $76.2 1.04
$100 $82.6 $81.2 1.02 $1000 $116 $130 0.89
• Significant consideration for inland utilities • RO system will have:
– Less brine flow – More concentrated brine – Anti-scalants in the brine
• Brine disposal cost will vary based on variety of factors including location, method, quality
56
Brine Disposal Method: When disposal of concentrate is difficult
• Examples of brine disposal methods: – Brine line/ocean outfall – Zero-liquid discharge (e.g., ponds, crystallizer) – Deep-well injection – Trucking
57
Brine Disposal Method: When disposal of concentrate is difficult
Impact of Brine Disposal Cost on Blend Cost Comparison (10 mgd with 20% salt reduction) Disposal Cost
($/ac-ft brine) MF-RO
($/ac-ft blend)
EDR ($/ac-ft blend)
MF-RO:EDR
$0 $79 $76 1.04 $100 $83 $82 1.01
$1,000 $116 $134 0.86 $10,000 $446 $659 0.68
58
Brine Disposal Method: When disposal of concentrate is difficult
Other Factors • Size of treatment facilities • Chemical price volatility • Tax implications of capital investment • Deviations from cost analysis assumptions
– Discount rate of 6% – Facility lifespan of 20 years – Electricity costs – Labor rate (average cost of $65/h)
59
Conclusions
60
Conclusions • Study first to provide full-scale cost comparison • Both technologies feasible for desalting recycled water • EDR less expensive (~4%) than MF/RO on salt removal
basis • MF/RO has higher capital costs, but less maintenance
labor required • Cost comparison close enough that site specific
impacts on cost must be considered – Seasonal usage will favor EDR (lower capital cost) – High brine disposal and/or land use cost will favor RO
61
Acknowledgements
City of San Diego
62