TECHNICAL PAPER 188

Increased Water Treatment Plant Capacity

at a Mexican Power Station Using Micro
Media Filters Ahead of RO and Replacing
EDI with Ion Exchange
RUSI KAPADIA, MICHAEL SHEEDY, DONALD SWAINE, Eco-Tec
Inc., Pickering, Ontario, PATRICIA M. SCROGGIN, Burns &
McDonnell, Kansas City, Missouri.

IWC-07-45

KEYWORDS: Ion exchange, compressed short bed, wastewater, filter, secondary water,
tertiary water, reverse osmosis (RO), electrodeionization (EDI).

ABSTRACT: This paper describes the process and equipment used to treat a Mexican
secondary water source in order to supply boiler feed make-up to a combined cycle power
plant. RO pretreatment is provided by a highly efficient micro media filter that produces a
filtrate with a turbidity of 0.1 – 0.2 NTU and an SDI value of 3 – 5. Following the RO, a
compressed-bed ion exchange system was supplied to replace an existing electrodeionization
polisher. The final product water produced has a conductivity less than 0.06 µS/cm.

the production of boiler water makeup

INTRODUCTION water. This is a gas-fired combined cycle
plant with a generating capacity of 600 MW
Recovery of wastewater for industrial use and has been in operation since 2002. The
presents many challenges for the design plant obtains its water (for both cooling and
and operation of water treatment boiler feed makeup) from municipal primary
equipment. The term wastewater can apply treatment lagoons/settling ponds. The main
to a range of water sources from municipal source of water entering the lagoons is
sewage to storm water runoff to industrial municipal sewage. Minor sources include
discharge water. As such, the water storm water runoff and industrial discharge
characteristics may have great variability water (both process and sewage).
and equipment used to treat the water must Secondary treatment of the water is
be robust and forgiving. Wastewater provided at the power plant itself. The
typically has high levels of dissolved solids, sewage treatment plant uses a biological
higher metals content than in natural water treatment process to first oxidize organic
sources and potentially high levels of matter and NH3 in an aerobic step and then
organic matter, both natural and man- remove nitrates formed by NH3 oxidation by
made. bacterial action under anaerobic conditions.
This paper describes a water treatment This eliminates biological contaminants and
system fed with secondary water used in reduces other contaminants in the water.
TECHNICAL PAPER 188

The water is then treated with lime, which Potassium mg/L as K 8.5
raises the pH of the water, causes the Fluoride mg/L as F 0.43
Nitrate mg/L as NO3 16.9
precipitation of dissolved minerals such as
Sodium mg/L as Na 230 – 330
calcium and magnesium and reduces the Chloride mg/L as Cl 250 – 350
overall dissolved solids content. The lime Manganese mg/L as Mn 0.2 – 1.0
addition takes place in a clarifier where the Iron mg/L as Fe 0.2 – 0.5
precipitates are removed. The water is then
treated with sulfuric acid to bring the water After extensive evaluation the system
to a neutral pH and disinfected by chlorine shown in Figure 1 was supplied by Eco-Tec
addition. The precipitated sludge is and consisted of these main elements:
thickened and dehydrated on a belt press to
produce a non-hazardous solid, which can Pretreatment using a single high
be sent to a landfill. efficiency micro media filter to
The clarified water is then suitable for use replace 4 existing sand filters.
as makeup water for the cooling towers. A Upgrade of existing Reverse
portion of the water requires further Osmosis capacity with new
treatment in order to obtain water that is membrane elements, valving and
suitable for use in a boiler and for other instruments.
plant water users. The original water Replacement of EDI system with a
treatment plant consisted of sand filtration, compressed short bed IX
cartridge filtration, two-pass reverse demineralizer.
osmosis and electrodeionization.
After three years of operation the plant Waste

needed additional capacity. Also, the plant

was experiencing operational difficulties. Filter Waste RO Waste

Reverse osmosis units fouled rapidly and So dium

Hypochlo rite Clean - In -
Pla ce
required frequent cleanings. The EDI Syste m

system was not consistently producing Clarified Wa te r

Coag ulan t So dium Bisulfite

water of sufficient quality. Feed

Sp ectru m Micro
Medi a Filter
Filtered Water
Tank
Cartridg e
Filters
Reverse Osmosis
Units
(3 x 5 0%)
Burns & McDonnell was hired to assist in (1 x 1 00% ) (2 x 5 0%)

evaluating improvement of the pre-filtration An tiscalen t

ahead of the RO, increased RO capacity Ba ckwash Pump

(2 x 1 00% )

and improved product water quality.

The clarified feed to the plant was found to Waste
Waste
have the following characteristics: Treatmen t
Neutral iza tio n
Syste m

Table 1: Feed Water Composition

Parameter Units Design Value To Poin t
DM Wa te r
Recoflo Trifl o
Demine ralization Pe rmeate
Stora ge Tank Unit Tank
or Range of Use
(1 x 1 00% )

TSS mg/L 5 – 10
Silica mg/L as SiO2 10 – 20
Acid
TDS mg/L as CaCO3 720 - 1030 Caustic

pH --- 7 – 7.5 gra vi ty

flow DM He ati ng
Alkalinity mg/L as CaCO3 140 – 200 Water Syste m

Conductivity μS/cm 1800 – 2400 Othe r Use

Phosphate mg/L as PO4 1–3

Hardness mg/L as CaCO3 220 – 300 Figure 1 - Schematic showing proposed main
Calcium mg/L as CaCO3 100 – 180 system elements
TECHNICAL PAPER 188

service flow rate of ~6 gal/min/ft2.

PRETREATMENT Backwashing of the existing system took ~2
hours to complete and was only done on
Secondary water sources like this cannot one train at any given time to ensure a
be used directly as feed to RO systems sufficient supply of RO feed. Performance
because they contain relatively high levels of this filtration system was poor and there
of suspended solids, organics, colloids and were frequent spikes in the solids content of
biological material even after clarification. A the filtrate. Cartridge filters which followed
tertiary treatment process of some sort is the media filters were overtaxed and solids
required to prevent fouling of the RO leaked through to the RO. This had an
membranes. Amongst the most commonly adverse effect on RO performance, and
recommended techniques for pretreatment increased the required frequency of
are: conventional depth media filtration; membrane cleaning.
microfiltration (MF); and ultrafiltration (UF)
1,2
.
A major advantage of depth media filtration
is that the cost of media replacement is
relatively low, since service life is long and
the media itself is inexpensive. It is also not
prone to fouling and generally is quite
robust. Together with a reasonable initial
capital cost, this explains the prominent
place that this pretreatment method has
held for many years. Nevertheless, MF and
UF membrane methods do provide a barrier
for suspended solids and ensure
Figure 2 - Previous Two Stage Media Filter Trains
consistently low levels of TSS to
downstream processes. The downside is
High Efficiency Micro Media Filter
that these processes tend to have higher
initial capital costs and operating costs for
The two-train system of primary and
membrane replacement are appreciable.
secondary media filters has been replaced
Also, in some cases, membrane
with a single "micro media" dual-media filter
pretreatment may just move the fouling
(Figure 3), which differs in a few key
problem upstream from the RO.
characteristics from typical dual media
The original system for tertiary treatment of
designs.
the secondary water supply consisted of
two trains of two-stage media filters (Figure
2). The primary sand filters had an effective
media size of 0.85 mm, and the secondary
filters had an effective media size of 0.65
mm. Coagulant (cationic polymer) was
added to the feed stream before the
primary filters. Sodium hypochlorite also
was dosed into the feed for disinfectant
purposes. The filter vessels had diameters
of 4.6 ft and generally were operated at a
TECHNICAL PAPER 188

micro media vessel which has replaced the

previously used two-train system is 5.5 ft in
diameter and has a design service flow rate
of 15 gal/min/ft2.
Unlike membrane filters, the performance of
a depth media filter will vary over the
duration of a service cycle. The previous
sand filters exhibited the classical ripening
effect typical of media filters at the
beginning of the cycle where the initial
filtrate quality was poor and gradually
improved as the cycle proceeded. To
ensure that only high quality filtrate was
passed downstream to the RO, it was
necessary to bleed the initial ~10 column
volumes of filtrate to waste over the first 2
hours of the service cycle. The micro media
filter, however has a much shorter ripening
period: suitable filtrate quality is obtained
after 2 column volumes (<10 minutes) of
Figure 3 - Current Single Micro Media Filter
the process cycle.
While the top media layer consists of To maximize the cleaning efficiency of the
coarse anthracite (effective size ~0.7 mm) backwash, the micro media filter operates
similar to that used in many conventional with a simultaneous air scour/backwash
dual media filters, the lower "micro media" cleaning cycle. Water is first drained to the
layer is a high density media with an top of the media. Air and water are then
effective size of less than 0.10 mm. This is passed simultaneously up through the
significantly finer than the fine media used media bed. The water flow expands the
in most designs which generally has an media and allows the air to agitate the
effective size of about 0.35 mm. media much more violently and uniformly
Since the lower layer removes the residual than with water alone. When the vessel has
quantity of suspended solids not retained in been refilled with water, the air is turned off
the upper layer it effectively defines the and a water backwash flushes dirt from the
filtration efficiency. The use of finer media filter in the usual manner. The cleaning
allows operation at a higher service flow cycle for the micro media filter is
rate while maintaining adequate filtrate approximately 15 minutes long.
quality. While the higher service flow and
finer media result in larger initial (i.e. clean) Filter Performance
pressure drops across the filter, the
pressure drops are manageable (typically Prior to installation of the full-scale micro
~22 psi compared to ~5 psi for a media filter, a performance comparison was
conventional filter). carried out between the existing sand filter
As a result of the higher flow rate provided trains and a pilot-scale micro media filter (6
by the micro media filter, the diameter of inch diameter column). The feed to the pilot
the filter vessel required to treat a given system was the same as that to the sand
flow of water can be reduced. The single filters, the feed stream being split after the
TECHNICAL PAPER 188

point of coagulant addition. 3.5

Figure 4 shows effluent turbidities over a 3.0

three day period for which the inconsistent
2.5
performance of the sand filters is apparent.

Turbidity (NTU)
Feed
The initial point of upset for the sand filters 2.0
Filtrate

(about midday on January 18) coincided 1.5

roughly with the time when one filter was 1.0

SDI15 <1
brought back into service after being 0.5
backwashed. Unfortunately, no SDI
0.0
measurements were available during the 0 1 2 3 4 5 6 7 8

period of comparison. It is also important to Service Step Time (hours)

note that the coagulant dosage was not 25

necessarily at the point required for optimal psi

performance. This is the probable reason 20

for the filtrate turbidities never falling below

psi, gal/min/ft2
15
~0.2 NTU.
gal/min/ft2
10
Feed
2.0 Pressure Drop
Filtrate (Pilot Micro Media Filter)
Flow Rate
1.8 Filtrate (Existing Media Filter Trains) 5
1.6

1.4
0
Turbidity (NTU)

1.2 0 1 2 3 4 5 6 7 8
Service Step Time (hours)
1.0

0.8
Figure 5 - Operating Data for Full-scale Micro Media
0.6

0.4
Filter.
0.2

0.0 REVERSE OSMOSIS SYSTEM

Jan-17 0:00 Jan-18 0:00 Jan-19 0:00 Jan-20 0:00

Prior to the upgrade, the existing reverse

Figure 4 - Filtrate Turbidities for Sand Filters and osmosis system was configured into two
Pilot Micro Media Filter.
trains. Each train contained three stages.
Operating data for a service cycle for the
Three vessels with five elements in the first
full-scale micro media filter is shown in
stage, two vessels with five elements in the
Figure 5. The service cycles are typically ~8
second and one vessel with five elements
hours long. Filtrate turbidities of ~0.1 NTU
in the final stage. Each train produced 14.2
are achieved and SDI1,5 values are <3. The
m3/hr. A sample of permeate from the RO
unit was being operated at a flow rate
taken just prior to the upgrade indicated a
below its design value of 15 gal/min/ft2 due
conductivity of 392 µS/cm. At the time of
to flow limitations of the secondary water
the analysis the RO membranes were
source.
approximately three years old. The poor
permeate quality was attributed in part to
leaky seals between the membrane
elements and membrane degradation due
to fouling of the membrane.
The RO system upgrade involved replacing
the existing membranes with higher surface
area membrane elements. This enabled the
TECHNICAL PAPER 188

existing arrangement of vessels and piping treatment plant, but chemicals are still
to be used to produce higher flow capacity. required and stored onsite for such things
The permeate flowrate per train was as cleaning RO membranes, feed water pH
increased to 28.4 m3/hr per train. In addition adjustment, anti-scalants, coagulants, and
some existing valving and instruments were lime softening.
replaced. Since EDI uses no regenerant chemicals
The original membrane configuration used they are not added into plant waste,
membrane elements with a 7% lower however the salts removed into the effluent
surface area and a flux rate (permeate flow stream from the concentrating compartment
per membrane surface area) of 7.5 of the stack technically would be classified
gal/ft2/day. The upgraded RO system as a waste and presumably would be
operates at a flux of 14 gal/ft2/day. The use identified as such under most
of a design flux rate of 14 gal/ft2/day is environmental operating permits. Other
based upon the RO element manufacturer’s concerns related to this technology include
guidelines for wastewater pre-treated with hardness scaling, leaks and the inability to
multi-media filtration or ultrafiltration. field service the stack assemblies.
Despite the increase in fluxes the The plant decided to replace the existing
membranes have experienced less fouling EDI cells and expand capacity using a
since commissioning. Once commissioned packed bed ion exchange system. Aside
the upgraded RO system produced from the more favorable economics, the
permeate with a conductivity ranging from principal technical considerations in arriving
30 - 70 µS/cm. at this decision were the rapid start-up after
Just less than half of the RO permeate is shut down provided by the packed bed
passed to the ion exchange process. The system, elimination of the second pass RO
balance is used elsewhere in the plant. requirement, and operating and maintaining
a single type of polishing system.
COMPRESSED SHORT BED ION
EXCHANGE SYSTEM Packed and Compressed Bed
Ion Exchange
The original system utilized EDI stacks.
This well established technology is typically In polishing applications resins are normally
used to treat double pass RO permeate. A used in mixed bed polishers that are either
detailed description of EDI with a regenerated on site or are shipped to
comparison to ion exchange can be found centralized off-site locations. For economic
in reference (3). The only concern reasons off site regeneration is limited to
expressed about the installed EDI feed streams with very low TDS. In this
equipment, beyond the requirement for case a design TDS in excess of 80 mg/L as
additional capacity was the slow and CaCO3 makes this an impractical option.
inconsistent restart after shutdown. Standard onsite mixed bed systems have a
The principal benefit of EDI is usually long and proven history of successful
identified as operation without the performance. However, relative to newer
consumption and storage of regenerant EDI and packed bed technologies operation
chemicals and the lack of a waste stream. and particularly regeneration can be much
Indeed, with the exception of some make- more complicated, particularly when resins
up salt, no regenerant chemicals are age. Many of these problems are related to
consumed by the EDI portion of the water resin separation and uniform re-mixing
TECHNICAL PAPER 188

before and after regeneration4. Another Demineralizer Design Basis

disadvantage of conventional mixed bed
technology is the relatively high regenerant The bid specification identified the
consumption and waste water production. composition of the RO permeate being fed
Packed resin bed technology reduces these to the demineralizer as follows:
problems by using separate beds of cation
and anion resin that are counter-currently
regenerated. Separating the resins
eliminates the separation and mixing Table 2: Demineralizer Feed Composition
problem. Counter-current regeneration Parameter Unit Value
minimizes regenerant consumption and Calcium ppm as CaCO3 4
allows production of high quality water by Chloride ppm as Cl 32
ensuring the highest degree of resin Silica ppm as SiO2 3
regeneration at the bottom of the column. pH 7.15
Thus, the cleanest resin is the last to Temperature Celsius 16 - 30
contact the product water ensuring the Bicarbonate ppm as CaCO3 30
lowest contaminant levels. Nitrate ppm as NO3 5.9
A further refinement to packed bed Sodium ppm as Na 35.91
technology is the compressed short bed Sulfate ppm as SO4 7
system5. Its most obvious feature is the Note that the average operating TDS is 30
short bed height of 3 or 6 inches. Other mg/L as CaCO3. The system was designed
features include the use of a much smaller to give a net product flowrate of just under
resin bead and complete elimination of 100 gpm. The target product water quality
internal freeboard. The smaller bead size is shown in the following table.
greatly improves the kinetics of the Table 3: Product Water Quality Target
exchange process, which allows operation Parameter Units Value
at higher flowrates and reduces the depth Conductivity µS/cm < 0.06
Total Silica as ppb <5
of the active exchange zone (permitting the
SiO2
use of such short columns). Eliminating
Copper as Cu ppb <2
freeboard and operating the resin in a
Sodium as Na ppb <5
slightly compressed state ensures that flow Sulphate as SO4 µg/L <3
is uniformly distributed (crucial for such Chloride as Cl µg/L <3
short beds) and that the resin position is TOC µg/L < 20
maintained during regeneration in order to pH 6-7
obtain full benefit of the counter-current
operation. System Description
One of the main concerns when using
packed bed systems is resin fouling. Since When using separate beds of cation and
resin is not backwashed during anion resin the limiting factor with regards
regeneration, any accumulated dirt will not to product water quality is generally sodium
be removed. Hence the need for highly leakage from the cation bed or from
efficient pretreatment. With this in mind residual caustic left on the anion resin after
treatment of RO permeate by packed bed regeneration. This becomes more
systems is an ideal application. problematic when trying to produce high
quality water (<0.1 µmho/cm) and with
higher feed TDS (>25 mg/L).
TECHNICAL PAPER 188

Given the feed composition and product concentrated regenerants are supplied to
water quality target a cation/anion bed metering pumps on the skid and injected
combination is not sufficient even with a inline to the regen water. Regenerant
compressed short bed system. In this case consumption and waste volumes are given
a third polishing bed of cation resin is below.
added to remove low levels of sodium. This Chemical Consumption
low level allows the use of a bed only 3” in Sulfuric Acid * 0.53 lbs/1000 gallons
depth that is operated at an even higher product
flux with very infrequent regeneration. Sodium 0.45 lbs/1000 gallons
A single train skid-mounted compressed Hydroxide * product
bed system with cation/anion/polishing Waste Volumes 17 gals/1000 gallons
cation beds was supplied and is shown in product
Figure 6. The main cation and anion beds * - denotes 100% weight basis
are only 30” in diameter and 6” in depth. These values are based upon treatment of
The final polishing cation bed is 24” in the 30mg- CaCO3 /l average feed TDS.
diameter and 3” deep. Net product flow is Even though this is a relatively high TDS for
100 gpm with a total system cycle time of a polishing application, regenerant
21 minutes. This includes the 5 minute fully consumption is small. Based on the above
automated regeneration sequence. Given numbers and assuming continuous system
the low sodium load to the polishing bed operation standard caustic and acid supply
regeneration is required only after 100 totes would last 1 and 2 months
cycles of the primary beds. respectively. This helps to reduce chemical
volumes stored on site and also reduces
the hazards associated with chemical
transfer and handling.

System Performance

The system was commissioned in August of

2006 and all performance objectives were
achieved. The equipment has since been in
continuous service without any problems.
Figure 7 shows the onstream conductivity
Figure 6 – Single train, skid-mounted, compressed profiles after the primary cation and anion
bed system. beds and then the final product water profile
To ensure < 5 ppb silica in the product the after the polishing bed, which confirms that
anion bed is regenerated with warm caustic the product water was <= 0.06 µmho/cm
at 60ºC. Since, the caustic regenerant target value. Feed conductivity was 50 – 76
volume is less than 10 gallons a small 3 ft µmho/cm. Product SiO2 levels were
diameter electrically heated tank is used to measured at less than 2ppb.
supply warm water for regeneration.
The only other significant auxiliaries are two
small 4 ft diameter tanks for combining the
regen wastes and trimming pH before
discharge. Day tanks for premixing of
regenerants are not required. The
TECHNICAL PAPER 188

Short Bed IX - Conductivity Profile REFERENCES

0.25
1. K. Sinha, "Prepared Discussion of
Conductivity (umho/cm)

0.2
Pretreatment Methods for Reverse
0.15
Primary Cond. Osmosis Systems", Proc.
0.1
Polisher Cond.
International Water Conference,
0.05 Pittsburgh, PA, Oct. 2001.
0
2. S. Gare, "Pretreatment Methods for
0 5 10 15 20 25 Reverse Osmosis Systems", Proc.
Percent Throughput (%)
International Water Conference,
Figure 7 – Onstream conductivity profiles Pittsburgh, PA, Oct. 2001.
3. Smith, J.H.; Hyde, B.; Crossen, M. –
SUMMARY “Short-Bed Deminaralization: An
Alternative to Electrodeionization”
An upgraded and expanded water Technical Paper presented at the
treatment plant was supplied to treat a Sixth International Conference on
secondary waste water stream from a Cycle Chemistry in Fossil Plants
Mexican municipal waste water treatment (EPRI), Columbus, Ohio, June 2000.
plant. A high purity product water with a 4. Smith, J.H.; Renouf, P.W.; Crossen,
conductivity <= 0.6 µmho/cm was produced M. – “50 Years in Separate Beds”
and supplied as boiler make-up water to a Technical Paper presented at the
600 MW combined cycle power plant. International Water Conference 1984
Tertiary treatment is provided by a high 5. Sheedy, M.; Kutzora, P. – “The Use
efficiency media filter that replaced existing of Short Bed Ion Exchange
sand filters and was found to be more Technology for the Product of High
economic than ultrafiltration. This filter Purity Water at WE Energies PPPP”
typically produced filtrate with a turbidity 0.1 Technical Paper presented at the
NTU and SDI values of <3. International Water Conference 2004
Primary demineralization is provided by a
single pass RO system using low fouling
composite membranes. Permeate quality
ranges from 30 – 75 µmho/cm when
treating feed water with a TDS of 700 –
1,000 mg/L.
Polishing is provided by a compressed
short bed ion exchange system, which
replaced existing EDI cells. The new
system is able to operate on a feed with a
much higher TDS eliminating the need for
the existing second pass RO. The problem
of slow EDI start-up after shut down was
also addressed. The system has now been
in service for over a year without any
problems.