Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ABTCP 2001
Jaime Staton
(Ecolochem International, Inc.)
Summary: This paper documents a successful reverse osmosis system upgrade from
traditional composite membranes to low energy, high surface area composite membranes
recently introduced into the marketplace. The paper gives specific attention to the
circumstances driving the upgrade, the site engineering issues needed to accommodate the
change, and the immediate and long term benefits. Data from the site shows that the upgrade
allowed for a 25% increase in system capacity with a negligible change in power consumption.
In addition, the paper compares performance characteristics of the two membrane types.
Keywords: membranes, reverse osmosis, low energy, low pressure, pulp and paper mill
Resumo: Este artigo descreve uma melhoria bem sucedida, executada em um sistema de
osmose reversa, mediante a substituição de suas membranas tradicionais por membranas de
baixa energia, e elevada superfície, recentemente introduzidas no mercado. Este artigo técnico
dedica especial atenção às circunstâncias que geraram a melhoria, aos assuntos relacionados
à engenharia do site, necessária para acomodar a alteração, e aos benefícios imediatos e de
longo prazo. Os dados do site mostram que a melhoria permitiu um acréscimo de 25% na
capacidade do sistema, com uma alteração despresível no consumo de energia elétrica. Além
disso, o artigo técnico compara as características de performance dos dois tipos de
membranas.
INTRODUCTION
Reverse osmosis (RO) has grown in its popularity as a means of reducing the ionic loading of
feed waters on downstream ion exchange (IX) demineralizers (DI). Using RO increases the run
times of the downstream demineralizers and reduces the frequency of regenerations and
chemicals needed for the system. The marriage of RO and IX has proven to be a natural step in
the ever evolving world of water treatment technology.
The new evolution is now focused directly on RO membrane technology itself, beginning with
the introduction of cellulose acetate (CA) membranes, followed by the concept of composite
membranes based on polyamide (PA) chemistry. CA membranes have historically been the
standard workhorse for RO systems being used in the water treatment industry. The use of PA
membranes has increased due to their higher performance guarantees versus CA membranes.
A recent market review predicts that membrane sales will continue to grow at a rate of 8%
1
through the year 2006.
The current trend in the membrane marketplace is the development of composite membranes
capable of reducing energy costs, while at the same time giving comparable salt rejections as
traditional composite membranes. These newer membranes are commonly marketed as “Low
Energy”, “Low Pressure”, or “Energy Saving” membranes. While standard 8” spiral wound
composite membranes have historically provided surface areas between 330 to 400 square
feet, low energy membranes typically start at 400 square feet and can go as high 440 square
feet. This increase in surface area, along with enhancements made to membrane structure and
chemistry, allow low energy membranes to achieve higher product flows than traditional
2
composite membranes, at manufacturer’s claims of 25-40% lower operating pressures.
Because both membranes share the same polyamide chemistry, these newer membranes
operate under the same physical and chemical constraints as standard composite membranes.
Table 1 lists the typical operating parameters of low energy membranes. Manufacturers of low
energy membranes include Dow, Hydranautics, Koch, and Osmonics.
The major driver in the this trend is cost savings. End users of high purity water treatment
equipment are constantly looking for ways to trim their bottom line. Applications best suited for
low energy membranes might be those where lower energy costs outweigh the expected
reduction in effluent quality. If the membrane performance can remain consistent over time,
then there are several advantages for using them:
As stated previously, this paper focuses on the upgrade of an existing RO system from standard
composite membranes to low energy composite membranes. The retrofit took place at a major
pulp and paper manufacturer in the eastern United States. Data taken before and after the
change out provides a comparison of the the two membrane types for parameters such as
effluent quality, flow, and pressure requirements. The paper also briefly examines the potential
cost savings claimed by membrane manufacturers by using low energy membranes versus
standard composite membranes.
CASE HISTORY
This eastern U.S. paper mill produces secondary fiber and both hard and soft wood pulp, which
is used in their liner board and sack product lines. For production the mill uses a large amount
of steam to operate their two black liquor recovery boilers and four power boilers. Prior to using
RO technology, approximately 60% of their condensate was returned as boiler feed water with
the remaining 40% make-up coming from on-site wells treated by the plants demineralizer
system. This system consisted of a three step ion exchange process using strong acid cations,
strong base anions and polishing mixed beds. In addition, the system contained a steam driven
vacuum decarbonator to reduce carbon dioxide (CO2) loading on the strong base anions. Boiler
make-up water usage averaged 1200 gpm.
The on-site well’s total dissolved solids (TDS) content is <600 parts per million (ppm),
characterized by low hardness (< 20 ppm as CaC03), high silica (20-35 ppm) and a turbidity
value of less than 1.0 NTU (nephelometric turbidity units). Silt density index (SDI)values
consistently measure less than 1.0 for a 15 minute test. SDI tests are routinely used to estimate
3
the RO fouling tendency of a water. Spiral wound membrane systems usually require feed
water with a silt density index (SDI) value of less than 5.0. A typical water analysis of the on-
site wells is given in Table 2.
Table 2 - Typical Raw Water Analysis
Cations (mg/L as CaCO3 ) Anions (mg/L as CaCO3)
Calcium 4 - 10 Bicarbonate 200 - 300
Magnesium 4-6 Chloride 20 - 30
Sodium 200 - 300 Nitrate 1-3
Potassium 9 - 10 Sulfate 2-3
Silica 20 - 35
TDS 450-600 ppm TOC 20 - 25
Conductivity 300-650 PH 7.5 - 8.5
µmhos
Several issues led to the mill’s decision to consider outsourcing RO equipment to supplement
their make-up water system. Prices for regenerant chemicals, particularly caustic, were
expected to continue their upward trend. Also during this time, plant operators were
regenerating up to 3 vessels per shift, and they could no longer count on the undersized waste
neutralization system to handle the large volume of regenerant waste. This led to periods of
boiler feed water shortages due to the inability to perform regenerations when needed. Another
issue of concern was the increasing age of some of the demineralizer equipment and the
anticipated high cost of replacement.
Late in 1997 an outsourced water treatment system was installed to provide 1200 gpm of RO
permeate to feed the plant’s existing three step demineralizers. This new system consisted of
ion exchange softening of the well water, followed by downstream reverse osmosis. A
converted salt cake silo was used to store the RO permeate. See Figure 1 for a process flow
diagram.
The reverse osmosis system consists of five individual RO units, each designed to produce 240
gpm, operating at a flux rate of 13.2 gallons per square foot of membrane surface per day
(GFD). Each 240 gpm RO unit consists of two 120 gpm trains, utilizing 3 x 2 x 1 reject staged
tube arrays with 6 membranes per tube. Standard 365 square foot composite membranes were
originally installed in the five RO units for a total membrane count of 360 elements. The
outsourced system had to meet the effluent guarantees listed in Table 3. A bypass line to the
silo was also installed to blend softened well water with RO permeate when make-up demands
exceeded 1200 gpm. Level controls in the silo automate the start and stop of the RO units, as
well as the opening of the bypass valve.
Figure 1 - System Process Flow Diagram
RO By-Pass Line
Reverse Osmosis
Well Water
RO Reject
The immediate result of the addition of RO to the make-up system was an 87% reduction in the
use of regenerant chemicals. This equated to only one regeneration per day versus three
4
regenerations per shift without RO pretreatment. The system operated without exception for
almost two years. During this time it became evident that water demands exceeded the 1200
gpm currently installed. The make-up water demand was expected to reach 1500 gpm.
Because of the increased demand for boiler make-up water, the RO bypass line began opening
on a regular basis. The blending of the softened well water with the RO permeate caused the
TDS of the ion exchange system feed to increase significantly. This directly impacted the
operating costs of the ion exchange demineralizer system. First, because of the increase in
alkalinity, the plant was forced to provide steam to operate the vacuum decarbonator to
minimize the impact on anion run lengths. The second impact was the shorter demineralizer
run lengths and the subsequent increase in regeneration frequency. This increase in operating
costs prompted the plant to consider upgrading the capacity of the existing RO system. The
decision on exactly how to upgrade was complicated by a lack of space in the RO building for
additional equipment as well as the plants desire to avoid major system changes.
SYSTEM MODIFICATIONS
Several scenarios were considered for the upgrade, including the installation of additional RO
units outside the RO building or replacing the current RO units in the building with larger units,
all of which required additional capital equipment and significant cost. Because of the excellent
quality of the feed water with respect to suspended solids, this system was considered an ideal
candidate for using low energy, high surface area membranes. The plant eventually decided
against the additional equipment and for the low energy membrane upgrade. Replacing the
existing membranes with new 400 square foot membranes and increasing flux rate to 15 gfd
would allow each RO unit to produce 300 gpm. The customer specifications listed earlier in
Table 3 would still apply, except for the flow rate specification, which could increase to 1500
gpm.
Over a three month period from March 1999 through May of 1999, each of the five RO units
were retrofitted with the new, low energy composite membranes. The expected increase in both
feed and permeate flows also required specific engineering modifications to the system. High
flow, low energy pumps were installed to replace the original high pressure RO feed pumps,
providing increased flow rates at lower pressures. The individual RO pump feed and discharge
piping was increased from 3” to 4” to accommodate the increase in feed water flow. To
compensate for the higher product flows, the common RO permeate line was upgraded from 6”
schedule 80 PVC to 8”, 316 stainless steel. The rest of the system remained unchanged.
Data had been gathered on the standard composite membranes for a period of approximately
seventeen months beginning from the job start up in November 1997 up until the three month
retrofit period beginning in March 1999. Data for the low energy membranes was then collected
after the retrofit, up until June of 2000, a period of approximately one year.
SALT REJECTION
Table 4 illustrates the average feed and permeate quality before and after the membrane
change out. The average feed conductivity of the well water has increased about 10% since
the RO system was first installed. The RO system loaded with standard PA membranes
averaged 75% recovery with a salt rejection rate of 97.4% based on an average permeate
conductivity of 11.5 µmhos and an average feed conductivity of 445 µmhos. The RO system
loaded with the low energy PA membranes averaged 73% recovery with a salt rejection rate of
97.0% based on an average permeate conductivity of 14.7 µmhos and an average feed
conductivity of 490 µmhos. The higher average recovery rate of system before the upgrade
was due to the higher demand for permeate during peak periods.
Comparing the in/out rejection rates based on conductivity, the standard PA membranes appear
to achieve better salt rejection than the low energy membranes by 0.4%. The graphs in
Figures 2 and 3 compare the normalized salt passage versus time for the RO system before
and after retrofit with low energy membranes. No noticeable increase in salt passage over the
first year of operation suggests good performance reliability.
2 0,8
% Salt Passage
1,8 0,6
1,6
0,4
1,4
1,2 0,2
1 0
03/30/1999
05/10/1999
06/28/1999
08/03/1999
09/13/1999
10/26/1999
11/30/1999
01/10/2000
02/15/2000
03/23/2000
05/15/2000
0,8
01/12/1998
02/17/1998
03/30/1998
06/15/1998
08/12/1998
09/29/1998
11/12/1998
01/14/1998
02/22/1998
04/05/1998
In this case, the change to low energy membranes from traditional membranes resulted in a
slight reduction in permeate quality, but not enough to fall below the customers specification of
94% rejection for the RO system. There has been no increase in the number of downstream ion
exchange regenerations since upgrading the system, in fact regenerations have declined to
once every three to five days as explained in the next section. The data does in fact support the
manufacturers claims of a slight reduction in quality.
PERMEATE FLOWS
The average permeate flow of the system containing standard membranes was 1100 gpm.
Since the upgrade to low energy membranes, the average permeate flow has increased by
16% to 1280 gpm. Figures 4 and 5 illustrate the permeate flow rates of both RO systems.
Figure 4 shows a decline in permeate production, when the mill’s demand was actually
increasing. This is directly attributable to the way in which the level control functioned. As
previously mentioned, a higher volume of softened water was bypassing the RO to meet
increasing plant water demands. It’s worth noting from Figure 5 that demand has peaked to
over 1500 gpm on several occasions. During these brief periods the RO recovery and flux rates
increased to meet demands. By installing more square footage of membrane surface area and
operating at a flux rate of 15 gfd, the system could meet the plants demand of 1500 gpm of RO
permeate. These changes allowed the total system capacity to increase 25% without the
addition of capital equipment.
GPM
1100 1300
1200
1050 1100
1000 1000
12/08/1997
02/02/1998
03/23/1998
06/08/1998
07/27/1998
09/17/1998
10/28/1998
12/23/1998
02/15/1999
05/17/1999
07/12/1999
09/07/1999
10/21/1999
12/13/1999
02/07/2000
03/13/2000
05/01/2000
The mill immediately realized benefits from the increased system capacity. First, the need for
softened water to by-pass the RO was eliminated. The upgraded RO system has been able to
keep up with the mill’s demand at all times. During times when the RO permeate was
supplemented with softened water, the TDS content of the ion exchange system increased
significantly. Since the alkalinity is converted to carbon dioxide by the strong acid cation resin,
the mill had to continue operation of the steam vacuum decarbonator in an effort to minimize the
impact on the strong base anion beds. When the IX system feed consists of RO permeate only,
the CO2 level after the cation beds is low enough to allow the plant to shut down the steam
vacuum decarbonator. This equates to a savings of approximately $250,000 a year in operating
costs. The second immediate benefit of eliminating the RO bypass was the effect on the
frequency of ion exchange regenerations and subsequent chemical usage. Regenerations
dropped from once per day, during times of supplemental bypass, to once every 3 to 5 days
when fed by RO permeate alone.
As stated earlier, each of the five individual RO units has two separate trains with two separate
feed pumps, totaling 10 arrays. The average output for these pumps before the upgrade was
146 gpm at 343 psi. The pressure drops across the system for the standard composite
membranes remained steady during the 16 months in service with an average pressure drop of
80 psi. Data taken since the membrane upgrade shows an average pump output of 175 gpm at
244 psi. The average pressure drop across the system has been 100 psi, 20 psi higher than
with standard membranes. The immediate effect of the low energy membranes was in the
average net drop in feed pressure of approximately 100 psi.
Figure 6 graphs the normalized pressure drop since startup of one of the individual RO units
with low energy membranes. This graph is representative of the other RO units as well. The
graph Figure 7 trends the primary pressure of the same RO unit since start up with the new
membranes. There appears to be some evidence of fouling according to the normalized
graphs. Most of the pressure gains have appeared on the first stage of the RO system which
might suggest blocking in the membrane spacers or brine channels due to silt, dirt, or bio-
fouling. The increase in primary pressure may also be explained by the upward trending of
permeate flow shown in Figure 5. More pressure is required to push flux rates higher to meet
demands.
180
85
170
80 160
75 150
03/30/1999
04/27/1999
06/01/1999
07/26/1999
09/22/1999
11/11/1999
01/17/2000
03/21/2000
04/24/2000
03/30/1999
05/17/1999
07/07/1999
09/02/1999
10/21/1999
11/30/1999
02/07/2000
03/27/2000
In order to increase membrane surface areas, membrane manufacturers have decreased the
thickness of membrane spacers and reduced the size of the brine channels. This could make
the low pressure membranes more prone to fouling by suspended material. The SDI values of
this well source have always been low. Other than the inherent filtering action of the ion
exchange softeners and 5 micron cartridge filters on the RO inlet, no pretreatment was
designed into the system. To date the RO system has never needed cleaning. More data is
needed before making a judgment on the fouling potential of the low energy membranes.
Table 5 illustrates a comparison of the energy demand between the standard and low energy
membranes. The average data reported for the RO system with standard membranes includes:
feed flow of 146 gpm at 242 psi, operation at 75% recovery, and permeate flow of 110 gpm.
Using this data and an electrical rate of $0.05 per kWh, the electrical cost to produce RO
permeate was $0.30 per thousand gallons.
The average data reported for the RO system with the low energy membranes includes: feed
flow of 175 gpm
at 244 psi, operation at 73% recovery, and permeate flow of 138 gpm. Using the same cost per
kWh, the new electrical cost was $0.22 per thousand gallons of RO permeate produced. This
represents a net electrical cost savings of 27% by upgrading to low energy membranes. Of
course, there are additional costs associated with the engineering changes and other factors
that would need to be included for a detailed and accurate comparison, all of which lie outside
the scope of this paper.
CONCLUSION
The results reported here demonstrate that low energy membranes can be an effective choice
for use on well waters that are low in turbidity and exhibit low SDI values. At increased flux
rates, there was only a slight reduction in salt rejection compared with standard composite
membranes. The salt rejection also appeared to remain stable during the first year of operation.
Additionally, the low energy membranes allowed higher product flow rates at significantly lower
operating pressures. This combination resulted in an increase in system capacity without an
increase in power consumption. In fact, there was a net savings in the production cost for RO
permeate by upgrading from standard composite membranes to low energy composite
membranes.
REFERENCES
1. M. Henley, “U.S. Membrane Market Continues Growth Pattern”, UltraPure Water, 17(4), pp.
12-14 (April 2000).
2. “Cost Savings Potential”, Technical Applications Bulletin, ESPA Membrane Elements, pp. 11,
Hydranautics, Oceanside, California (July 1995).
3. Annual Book of ASTM Standards, “Standard Test Method for Silt Density Index (SDI) of
Water,” Volume 11.01, Section 11, (1998), Method D 4189-95.
4. C.R. Istre, W.R. Temple, “Outsource or Capital - Financing a Water Treatment Project,”
International Practical Industrial Water Treatment Technology Conference & Exhibition,
Houston, TX, (Jan. 1999), pp. 9-12.