EMERGING TECHNOLOGY REPORT
CROSS-FLOW PERVAPORATION SYSTEM
FOR REMOVAL OF VOCs FROM
CONTAMINATED WASTEWATER
by
Wastewater Technology Centre
Burlington, Ontario
CR815788-02
Project Officer
John Martin
Technology Demonstration Division
Risk Reduction Engineering Laboratory
Cincinnati, Ohio 45268
RISK REDUCTION ENGINEERING LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OH 45268
-------
DISCLAIMER
The information in this document has been funded wholly or in part by the U.S.
Environmental Protection Agency (EPA) under Cooperative Agreement CR-815788-02, to
Zenon Environmental Inc. It has been subjected to the Agency's peer and administrative
reviews and it has been approved for publication as an EPA document. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
i i
-------
FOREWORD
Today's rapidly developing and changing technologies and industrial products and
practices frequently carry with them the increased generation of materials that, if improperly
dealt with, can threaten both public health and the environment. The U.S. Environmental
Protection Agency (EPA) is charged by Congress with protecting the nation's land, air, and
water resources. Under a mandate of national environmental laws, the agency strives to
formulate and implement actions leading to a compatible balance between human activities
and the ability of natural systems to support and nurture life. These laws direct EPA to
perform research to define our environmental problems, measure the impacts, and search for
solutions.
The EPA Risk Reduction Engineering Laboratory is responsible for planning,
implementing, and managing research, development, and demonstration programs to provide
an authoritative, defensible engineering basis in support of the policies, programs, and
regulations of the EPA with respect to drinking water, wastewater, pesticides, toxic
substances, solid and hazardous wastes, and Superfund-related activities. This publication is
one of the products of that research and provides a vital communications link between the
researcher and the user community.
The primary purpose of this guide is to provide standard guidance for designing and
implementing a biodegradation treatability study in support of remedy selection testing.
Additionally, it describes a three-tiered approach that consists of 1) remedy screening testing,
2) remedy selection testing, and 3) remedial design/remedial action testing. It also presents a
guide for conducting treatability studies in a systematic and stepwise fashion for determination
of the effectiveness of biodegradation in remediating a site regulated under the Comprehensive
Environmental Response, Compensation, and Liability Act. The intended audience for this
guide includes Remedial Project Managers, On-Scene coordinators, Potentially Responsible
Parties, Consultants, Contractors, and Technology Vendors.
E. Timothy Oppelt, Director
Risk Reduction Engineering Laboratory
111
-------
ABSTRACT
The U.S. Federal Register of April 17, 1987, contains a list of priority pollutants found
at Superfund Sites. About half of these are volatile organic compounds (VOC's), which are
known to be toxic and/or carcinogenic in nature. Pervaporation is a membrane technology
utilizing a dense non-porous polymeric film to separate the contaminated water from a
vacuum source. A membrane is used that preferentially partitions the VOC organic phase
used in this test. This process has proven to be an alternative to conventional technology
because it removes the amount of VOC's without requiring any pre/post treatments.
Pervaporation is a cost-effective method of removing VOC's.
IV
-------
TABLE OF CONTENTS
Disclaimer i i
Foreword iii
Abstract iv
Figures vi
Tables ~ vi
Executive Summary 1
1.0 INTRODUCTION 2
2.0 BACKGROUND 3
2.1 Volatile Organic Compounds 3
2.2 Pervaporation 3
2.3 Activated Carbon and Air Stripping 6
2.3.1 Activated Carbon 6
2.3.2 Air Stripping 7
2.4 Relative Costs of Technologies for VOC Removal 8
2.5 Previous Pervaporation Work for Removal of Volatile Organic Compounds • 8
2.6 Importance of Liquid Film Resistance 10
3.0 PROJECT OBJECTIVES 11
4.0 EQUIPMENT DEVELOPMENT 11
4.1 TheMembrane 11
4.2 TheModule 13
4.3 TheSystem 14
5.0 QUALIFICATION TESTING 15
5.1 Feed Pressure Loss 15
5.2 Pervaporation Results (Bench Scale) 19
5.2.1 Pervaporation Results 19
5.2.2 Quality Assurance 21
5.3 Effect of Major Variables 25
5.3.1 Effect of Reynold Number 25
5.3.2 Effect of Feed Temperature 26
5.3.3 Effect of Different VOCs 28
5.3.4 Permeate Pressure 30
5.3.5 Effect of Feed Temperatures for Rough Vacuum Operation 31
5.4 Pilot Scale Operation 35
6.0 PROCESS OPTIMIZATION 36
6.1 Operating Conditions and Equipment 36
6.2 Case Study 36
6.3 Technical and Economic Analysis 38
7.0 CONCLUSIONS AND RECOMMENDATIONS 39
REFERENCES -.4 1
APPENDIX A 43
APPENDIXB 75
-------
FIGURES
2.1 Typical VOC Concentration Gradient from Membrane to Vacuum for
Pervaporation Operation 5
2.2 Schematic of the Pervaporation Process 5
4.5 WTC Bench-Scale Pervaporation System 16
4.6 Pervaporation Pilot System 17
5.1 Pressure Drop Through Array of 4800 Fibers ; . ... 19
5.2 Enhanced VOC Removal Promoted by Increase in Reynolds Number 26
5.3 Water Permeability Increased by Increasing Operating Temperature 27
5.4 VOC Removal for Toluene, TCE and EDC @ 25 Through 35°C Using a
Transverse Flow Pervaporation Module Operating Under High Vacuum and
a Water Side Reynolds Number of 40 to 60 29
5.5 Separation Factors for VOC Removal for Toluene, TCE and EDC
@ 25 - 35°C Using a Transverse Flow Pervaporation Module Operating
Under High Vacuum and a Water Side Reynolds Number of 40 to 60 30
5.6 Reduction in Water Flux with Increase in Module Permeate Pressure 32
5.7 Reduction in VOC Flux with Increase in Module Permeate Pressure 32
5.8 Condenser Pressure Estimated as a Function of Condenser Temperature
for a Saturated Toluene/Water Liquid 33
5.9 Increased VOC Removal at High Temperature for Rough Vacuum Operation .... 33
TABLES
2.1 Volatile Organic Compounds List of Priority Pollutants 4
2.2 Summary of Pervaporation Results for Organic Compound Removal
from Water 9
4.1 Description of Modules and Fibers Used for Testing and Determining
Hydrodynamic Conditions Applicable for a Transverse Flow 14
5.1 Coefficients for Estimation of Pressure Drop in Transverse Flow of
Water Across a Bank of Fibers 18
5.2 Pervaporation Test Results for Bench Scale Program 20
5.3 Reduced Analytical and System Data for Pervaporation Testing 22
5.4 Quality Assurance Precision and Completeness Results 25
5.5 Organic Compounds Considered for Removal by Pervaporation at
Various Operating Temperatures 29
5.6 Operating and System Parameters for Pilot Testing of a
Pervaporation Module 36
6.1 Benefits of Operation of Pervaporation System at Optimum Conditions 37
VI
-------
EXECUTIVE SUMMARY
Water contaminated with volatile organic compounds is encountered throughout industry and in
many groundwater and site remediation applications. Conventional technologies such as air
stripping and activated carbon treatment do not always provide a complete and economic solution
for some of these wastewater applications, Previous work has demonstrated that pcrvaporation is
a potentially suitable remediation method for such applications. The primary objectives of this
project have been to develop an improved membrane and module design to make pervaporation a
more cost-effective method of removing volatile organic compounds from contaminated water,
and to compare the improved pervaporation module and membrane design to other remediation
technologies as well as other pcrvaporation module and membrane designs for the removal of
organics from contaminated water.
Improved modules and membranes were developed, and a system was designed to test these
pervaporation modules. Testing was carried out by U.S. EPA accepted methods. Testing
confirmed that the transverse flow pervaporation module using a thick membrane provided
improved performance. In addition, important variables such as Reynolds Number, operating
temperature, permeate pressure and organic volatility were considered in pervaporation testing and
the effect of such variables were quantified at bench-scale. These results confirmed the validity of
existing models for predicting pressure drop through membrane modules and, for estimating the
rate of removal of organics from water by pervaporation. Removal rates and selectivities were
higher in this work using a transverse flow pervaporatibn module than reported elsewhere for
conventional modules. The bench-scale results were also verified at pilot-scale using a transverse
flow pervaporation module with 0.5 m2 of surface area
Models for mass transfer and pressure drop in a closely packed array of hollow fibres were used to
carry out sensitivity and optimization analyses. These analyses identified optimum operating
conditions for pervaporation operation. Some of these results were reported in Environmental
Progress (Lipski and C6t6, 1990). Optimization indicated that pervaporation operation with
transverse flow modules would be most effective using thick membranes at higher Reynolds
Numbers and higher operating temperatures
-------
1 QINTRODUCTION
This is the final report for the unsolicited proposal entitled "Development and Evaluation of a
Cross-Plow Pervaporation System for Removal of Volatile Organics from Contaminated Water",
Contract #09SE.KE405-8-6385. This work was funded by Environment Canada through the
Department of Supplies and Services.
The purpose of this work was to develop a cost effective pervaporation membrane module and
system to remove volatile organic compounds from contaminated water. Pervaporation has the
potential of replacing conventional technologies such as activated carbon adsorption and air
stripping.
The project was broken down into 8 tasks:
1. Membrane Requirements 5. Qualification Testing
2. Module Requirements 6. System Optimization
3. System Requirements 7. System Evaluation
4. System Construction 8. FinalReport
Amendment #1 dated January 18.1991 was to; 1) redefined the work statement of Task 5 to allow
for modification of existing test equipment rather than building new equipment, and 2) eliminated
field work from Task 7.
This project was also supported under the US Environmental Protection Agency SITE Emerging
Technology Program. One of the requirements of EPA was the development of a Quality
Assurance Project Plan that served as a guideline to obtain reliable experimental information. A
Quality Assurance section (Section 5.2.2 in this report) discussed the results of this Quality
AssuranceProgram
This report is organized in 7 sections. In Section 2, relevant background information is presented
and supports the statement of project objectives in Section 3. The pervaporation equipment
developed in the project is described in Section 4. Testing results are presented in Section 5 while
ways to optimize the process are outlined in Section 6. Recommendations and conclusions are
presented in Section 7.
Several publications have been produced in the course of this project. Copies of these are included
in the Appendices.
-------
2.0 BACKGROUND
2 . 1 Volatile Organic Compounds
Volatile organic compounds (VOCs) are common contaminants in wastewater, leachate and
contaminated groundwater. About half of the 129 US EPA priority pollutants are VOCs and are
known to be toxic and/or carcinogenic. VOCs are emitted in large quantities (1,600,000 to
5,000,000 metric tons per year) from waste treatment, storage and disposal facilities (Shen slaL
1988). VOCsare also present at abandoned industrial sites. The U.S. Federal Register of April
17, 1987 contains a list of priority pollutants found at Super-fund sites. A number of these
contaminants can be removed by pervaporation. These are presented in Table 2.1.
2.2 Pervaporation
Pervaporation has been considered (Bran, 1981; Eustache & Histi, 1981; Nguyen &Nobe, 1987)
as an alternative technology for removal of volatile organic compounds from contaminated water.
The contaminated water may be an industrial process water, groundwater or leachate.
Pervaporation is a membrane technology utilizing a dense non-porous polymeric film to separate
the contaminated water from a vacuum source (Figure 2.1). The volatile organic compounds
contained in the liquid phase are adsorbed onto the membrane and diffuse through to the other side
where they are drawn off by a vacuum. A membrane is used that preferentially partitions the VOC
from the water (much like an organic phase used in extracting organics from water samples in
liquid/liquid extraction). For water treatment applications, the membrane is made of an
organophilic polymer such as silicone rubber which exhibits good permeability for the organic
compounds while allowing very limited passage of water. A typical VOC concentration gradient
across the membrane is shown schematically in Figure 2.1. Although permeability through
silicone rubber may be four times higher for water than it is for VOCs, the preferential partitioning
of VOCs at the membrane/liquid interface provides an overall enrichment of VOC on the permeate
side of the membrane. Most organic compounds are concentrated in the permeate by orders of
magnitude compared to the aqueous waste. The organics and some water which passes through
the membrane are condensed, The condensed permeate often separates into an aqueous and an
organic phase, offering industrial applications the possibility of recovering the organic fraction. A
continuous pervaporation process is illustrated schematically in Figure 2.2.
-------
Table 2.1: Volatile Organic Compounds from the Federal Register
List of Priority Pollutants (April 17»19i7)
Contaminant
Priority Group 1
Chloroform
Benzene
Vinylchloride
DieMorometttane
Priority Group 2
Carbon tetrachloride
Chloroe thane
Bromodichloromethane
Dichloroethylene (1,1)
Dichloropropane (1,2)
TricMoroethane (1»1»2)
Tetrachloroelhane (1,1,2,2)
Toluene
Priority Group 3
TricMoroetfaane (1,1.)
CMorome thane
Bromofona
Dichloroethane( 1,1)
Ethyl benzene
Acrolein
Aorylonitrile
CMorobenzene
Chlocodibromomethane
Dichloroethene (trans)
Priority Group 4
Bromomethane
Carboa disulfide
TricWorofluoromethane
Dichlorofluoromethane
DIchlorobenzene (1,2) (~p)
Diox^aae (1,4)
DIchlorobenzene (1,3) (~m)
CASf
67663
71432
75014
75092
79016
106647
127184
56235
75003
75274
75354
78875
79005
79345
107062
108883
542881
71556
74873
75252
75354
100414
107028
107131
108907
124481
158606
74839
75150
75694
75718
95501
123011
541731
Vapor Pressure
@ 25°C law HgJ
208
95
2660
438
75
1
19
113
1200
59
630
40
25
7
82
30
30
123
3830
6
234
10
244
114
12
IS
200
1250
366
796
5000
2
37
2
-------
ft)
I"
Water Containing Membrane
VOC
Vacuum
Figure 2.1 - Typical VOC Concentration Gradient from Membrane to Vacuum for
Pervaporation Operation
Aqueous waste
Pervaporation module Treated effluent
r
Permeate
Feedpump
Membrane
Condenser
Vacuum pump
Organic liquid
Figure 2.2 - Schematic of the Pervaporation Process
-------
There are three companies that have developed pervaporation membrane modules for the removal
of organic compounds from water : GFT (Germany) with a plate-and-frame module, GKSS
(Germany) with a cassette type module, and Membrane Technology and Research (CA, USA)
with a spiral-wound module. At this point, these membranes and modules are under evaluation in
pilot- studies; no full scale application has been reported These systems are however not suitable
for applications involving low concentrations of VOCs which are typical of contaminated
groundwater due to poor economics and technical limitations (Lipski and C6t6,1990).
2 . 3 Activated Carbon and Air Stripping
Current VOC treatment methods generally utilize activated carbon adsorption and/or stripping to
remove low concentrations of organic contaminants from water. These methods have been used as
a basis for comparison to evaluate the potential of pervaporation for similar applications. Other
technologies which can remove VOCs include biological treatment and liquid or gas phase
oxidation (e.g. W-ozone). However, these technologies are not as widely used as carbon
adsorption and shipping for VOC removal, therefore they have not been considered in detail in this
report. The following driving forces exist for the replacement of activated carbon and stripping:
1) elimination of carbon disposal/regeneration costs;
2) elimination of air emissions from the stripping processes.
A brief comparison of these technologies vs. pervaporation is presented in the following sub-
section.
2.3.1 Activated Caibon
Activated carbon offers very high activation energies for most VOCs. Activation energy is defined
as the amount of energy required to release the VOC from the active material. For most VOCs on
carbon, this high activation energy requires heat (either direct fired or from steam) to release the
VOCs and regenerate the carbon. On the other hand, pervaporation takes advantage of the lower
activation energy offered by organophilic polymers compared to carbon by offering continuous
release of the VOCs to vacuum. Because of its high activation energy, activated carbon can not be
regenerated by vacuum
In activated carbon, organic compounds compete for adsorption sites and therefore the removal
efficiency decreases as these sites become saturated Certain compounds, such as ethylene
dichloride (EDQand methylene chloride although adsorbed by activated carbon, are quite often
"displaced by" other organic compounds which have higher activation energies for the activated
-------
carbon, Unlike carbon adsorption, there is no competition between organic compounds in
pervaporation. Compounds which diffuse through the polymer membrane are continuously
removed on the vacuum side and can not be released back into the aqueous stream.
Carbon adsorption is cost-effective for low concentration applications but becomes expensive at
higher concentrations because spent carbon must be disposed of or regenerated more frequently.
In addition, the effectiveness of carbon declines each time it is regenerated, In pervaporation,
since there is continuous release of the VOCs to the vacuum, the membrane never becomes
saturated and therefore never needs to be regenerated. In pervaporation, the permeate, which is
primarily organic liquid, must be disposed of as a hazardous waste if it cannot be reused in an
industrial process stream. This volume is, however, much reduced compared to the initial
wastewater allowing for more economical off-site transport to an approved hazardous waste
incinerator for destruction. Likewise, as carbon is regenerated, organic liquids recovered from
regeneration must be disposed of as a hazardous waste if not suitable for reuse. Finally,
pervaporation does not consume reagents or exhaustible sorbents. As landfill costs increase for
disposal of activated carbon (considered a hazardous waste), carbon will tend to be used only in
applications where other technologies prove to be ineffective.
2.3.2 Air Stripping
Air stripping is limited to the removal of compounds that preferentially partition into air compared
to water (i.e. high Henry's law constant). Furthermore, water containing dissolved solids often
promotes fouling of stripping columns due to iron oxidation and/or carbonate precipitation,
reducing process efficiency and resulting in increased maintenance costs. In pervaporation,
fouling is minimized because air is not added to the water being treated
In addition to the above limitations, unless the off-gas is treated, air stripping merely turns a water
pollution problem into an air pollution problem. A recent survey of 177 air stripping installations
at remedial sites in the U.S. showed that only 17 had any off-gas treatment, and for most of these,
data was not available on the efficiency of the process (Radian Corporation, 1987). It is
anticipated that the U.S. Clean Air Act will make off-gas treatment compulsory, greatly increasing
the complexity and cost of air stripping. The most popular method for off-gas treatment is carbon
adsorption, which has the same limitations described above for water treatment. In contrast, for
pervaporation, the organic compounds permeating through the membrane are contained by
condensation, thus providing an opportunity for the recovered organic phase to be reused
-------
2.4 Relative Costs of Technologies for VQC Removal
Typical treatment costs for air stripping and activated carbon were presented in a recent publication
(Lipski and C6t6,1990). This case study considered a 167 litres/min. system which contained 10
ppm trichloroethylene (TCE) which was to be reduced by 99%. Air stripping alone would cost
SO.lO/m3. The cost of combining air stripping with granular activated carbon (GAC) in various
configurations, varied from $ 0.40/m3 for air stripping and vapour-phase GAC, to $ 0.80/m^ for
liquid phase GAC alone. These costs include regeneration or disposal of activated carbon. By
comparison, the cost of pervaporation would be in the order of $ 0.5fi/nr using the membrane and
module design developed in this project. This cost was estimated from energy requirements and
amortization of assembly costs and component costs and indicates that pervaporation is cost
competitive with existing technologies.
Compared to competing technologies, pervaporation features the following benefits :
.Process is completely enclosed thereby minirnizing direct and fugitive emissions;
. suitable for concentrations ranging from ppb to g/L;
. no requirement for chemicals or adsorbents;
. systems are compact, modular and easily transportable;
. low operating costs;
. opportunity for recovering concentrated organics for recycle/reuse.
2.5 Previous Pervaporation Work f o r Removal of Volatile'Organic C o -
There has been some prior work in the removal of VOCs from water by pervaporation. Examples
of organic compounds which have been effectively separated from water using different
'membranes are presented in Table 2.2. The separation factor, a, which is a measure of the
increase in the concentration of the VOC in the permeate relative to the feed concentration, varies
significantly in the range 80 to 21,500. Use of mass transfer coefficients (MTCs) is a convenient
means used to describe the relative effectiveness of different membranes in terms of rate of
removal of VOCs from water.
Separation factors for all the systems are high and provide a measure of enrichment of the VOCs in
the permeate over the feed. In fact, most of the systems show separation factors that are high
enough to allow for collection of a super-saturated permeate which undergoes phase separation.
-------
Table 2.2 Summary of Pervaporation Results for Organic
Compound Removal from Water
Reference
Membrane
Thickness WaterFlus Ccmpound Separation MT Coeff.
Factor
<*•) (o/^/h) (*) 0*/»>
Eustache & Histi, 1961
Plane polyester coatad with silicone 100
rUbber (Rhone Poulenc TS605)
27 Ben-ETCi
chlorofom
Dtchlororoethane
Vinyl chlorid.
11100 N/A
6800 N/A
4300 N/A
9000 N/A
Brun, 1981
Nguyen & Nobe, 1987
Pswne, 1984
Butadiene styrene/acrylonitrile
Silicone rubber fiber
Dow Corning
Plane polysulfone coated with
s i 1 i con ntber
Plane polysulfone coated with
• silicom rubber
Silicone rtUaa^fiber
Dow Corning
Radroff & Lipski, 1988 Polyethylene
195 1.7 chloroform
165
0.8
15
165
50
13
530
160
14
40
14
chloroform
IraKMtthar*
Dichloromethane
Chloroform
Tr Ich I orocthy l«nt
Chlotofora
Trichlomethylum
21500 10
93QO 34
720026
4760 17
200
80
630
355
29
12
29
16
Tr
*ft«r«« obt«tr*d for ttynolds maber* varying from 2 to 200
Preferential permeability plays a minor role in the separation of most of the
compounds from water by pervaporation. As illustrated in Figure 2.1, water may
be preferentially permeable compared to the VOC, yet the membrane can demonstrate
excellent separation towards the VOC. The-differences in gas pemeabilities of
most VOCs range by no more than a factor of ten, with water being slightly more
pemeable, through silicone, than most VOCs. Other polymeric materials exist
which have demonstrated higher permeabilities towards VDCs than water
(Nijhuis,1990).
Preferential partitioning from the water to the membrane surface plays the
critical role in defining the effectiveness of pervaporation for the removal
of a certain VOC from a particular water stream. Henry's Law constant is a
useful parameter for estimating partitioning of a VOC from water.
Octanol/water partition coefficients also offer good indication of
partitioning from the water to the membrane surface. Henry's Law constants
are however more readily available in the literature.
9
-------
The preliminary work in pervaporadon was sufficient to demonstrate that this technology is
applicable to VOC removal from contaminated water. However, it soon became clear that the
module design and consequently the liquid film resistance (LFR) represented the rate controlling
mechanism (CM and Lipski, 1988) in this pervaporation application.
2.6 Importance of Liquid Film Resistance
Several groups, Yang and Cussler (1986) and L£vSque (1928), have demonstrated the importance
of liquid film resistance (LFR) in heat and mass transfer applications. In pervaporation, C6t6 and
Lipski (1988) and Nijhuis (1990) were able to show that, in most instances, removal of the VOOs
from water is limited by LFR for mass transfer. It has been demonstrated that although chlorinated
organic compounds have a very high affinity for silicone rubber compared to water, LFR is often
rate controlling and reduces the rate at which organics may be removed from the water. In fact,
Nijhuis demonstrated that separation would be increased by 3 or 4 times (for either toluene or
trichloroethylene with 60 - 240 \im thick silicone rubber) if there was no LFR.
The relative magnitude of the LFR compared to the membrane resistance can however be reduced
by using a thicker membrane to achieve the higher separation factor. The thicker membrane
typically offers reduced water flux with no significant decrease in VOC removal. Improving
hydrodynamic conditions (as described below) also increases the separation factor and has the
added benefit of improved VOC removal. Since there is a practical ceiling to improvement of
hydrodynamic conditions, it is apparent that membrane thickness should also be optimized for a
given hydrodynamic condition in order to provide high VOC removal and good selectivity.
In addition to improving the selectivity, using a thick membrane to reduce the water flux, without
sacrificing organic removal, has operational advantages. Reducing water flux (quite often, the
major permeate component) reduces vapour handling requirements and condensing duties
(estimated to be one of the highest sources of energy consumption in typical pervaporation
systems).
To overcome hydrodynamic limitations, Lipski and C6t6 (1990) were able to reduce liquid film
resistance and significantly increase removal of the target compound by using a transverse flow
module. Furthermore, improving hydrodynamic conditions, which reduced the liquid fihn
resistance, improved the removal of the organic compound even with a thicker membrane which
resulted in reduced water flux. Increasing the rate of VOC removal while significantly reducing
the water flux has significant advantages due to reduced vapour handling requirements and
operating costs. The impact of these effects on treatment cost was investigated in a sensitivity
analysis. It was concluded that a system utilizing a transverse flow module design offered the
most cost effective technique for VOC removal by pervaporation
-------
J .
0 PROJECT OB.TECTIVES
The principal objective of this project was to develop an improved pervaporation treatment process
to remove volatile organic compounds from waste streams. Secondary objectives include:
1) Development of hollow fibre membranes with the active polymer coated on the outside.
2) Optimization of the membrane thickness and hydrodynamic conditions to maximize process
efficiency.
3) Development of a transversal-flow prototype module with improved mass transfer
characteristics.
4) Testing of the pilot-plant with several VOC contaminated wastewater solutions to determine the
performance and provide scale-up data.
5) Carry out technical and economic analysis of the process compared to conventional VOC
removal processes.
EQUIPMENT DEVELOPMENT
The three major tasks of the project involved development of the membrane, the module which
houses the membrane and, the pervaporation system which enables the module'to perform
effectively are summarized in the following subsections:
The Membrane
The objective of membrane development work was to develop and test a hollow fibre with a
pervaporation membrane on the outside surface of a supporting layer. This development work
was aimed at defining a method for making a membrane on a fibre support to allow subsequent
testing to be carried out with liquid flow transverse to the fibre axis.
A number of criteria were used to define the requirements of the membrane and its supporting
layer. The membrane and support material had to be resistant to a wide variety of solvents that
may be encountered in typical contaminated groundwaters. These materials also had to be suitable
for testing at elevated temperatures up to 60°C. The support fibre needed sufficient strength to
allow pervaporation testing with high liquid velocities orthogonal to the fibre direction. Previous
11
-------
work indicated that a thick (at least 30 Jim) silicone membrane would be most suitable for
pervaporation testing. Other polymeric membranes, such as polyethylene or polypropylene, etc.,
were also identified as being suitable for use as pervaporation membranes. Fiially, the membrane
material used to coat the fibres had to be suitable for potting in order to allow the fibres to be sealed
in a water-tight module.
Several options were available to allow testing of pervaporation fibres with transverse liquid flow.
Pure silicone fibres were considered since they were commercially available and used in previous
work Although these fibres are readily available as a source for pervaporation membranes, they
are relatively expensive. Silicone fibres also require a support material to provide a means of
handling and orienting the fibres into the desired configuration.
Celgard fibre, available from Hoescbst Celanese Corporation, was retained as the support material
as it provided solvent resistance and was temperature stable to 70°C. This was a microporous
polypropylene fibre with a wall thickness of 30 Jim The microporous material had a benefit in
this application as it allowed the membrane to anchor itself into the support. Silicone rubber was
chosen as a membrane material since it is available in a two component base and catalyst which
provided easy polymer fabrication without need for any complex polymer fabrication equipment.
Potting is a critical step in the module manufacturing process since silicone rubber can not be
potted with available epoxies, membrane coating after module construction was considered first
Coating was first attempted by pouring silicone over a potted module army of fibres, then allowing
the silicone to drain. The silicone material proved to be too viscous and did not allow sufficient
drainage. Coating was then attempted on a potted module array, but this time by filling the
micropores of the support material from the inside of the hollow fibre. Since the pores were
relatively large and allowed passage of the silicone, the silicone was pushed (at 25 psig) through
the pores to form a layer of silicone on the outside of the fibre. The bore of the fibre was then
purged with gas to remove the silicone prior to cross-linking. The coating thickness on the outside
of the fibre was controlled by adjusting coating time. Coating thickness on the inside of the fibre
was controlled by adjusting purge time. This method proved effective for producing membranes
from 30 [Urn thick (the thickness of the Celgard fibre wall) up to 150 pum thick. Membrane
thickness was also controlled by addition of solvent (such as pentane) to reduce membrane
thickness upon application.
12
-------
4.2 The Module
The objective of this phase was to develop a transverse flow, hollow fibre, lab scale prototype
module.
The module was required to'contain capillary fibres (less than i mm diameter
which were suitable for transverse flow operation. The fibres had to be
arranged in a pattern such that liquid flow outside the fibres js orthogonal
to fibre direction, and channelling is minimized by the fibres which are
pulled taut and spaced in an ordered, repeating matrix.
Several options provided means of satisfying these requirements. Weaving fibres to form a mesh
of membrane fibres amongst guide fibres (made of non-membrane material) was considered as a
method of orienting fibres to provide a form suitable for manipulation and subsequent potting.
Direct fibre placement onto a grid network was also considered for module preparation. Weaving
of fibres was contracted out to determine feasibility for fibre preparation as a part of module
production. Arrays of woven fibres did not however possess well defined and repetitive spaces
between fibres and would be prone to fouling and channelling of liquid flow.
Direct placement of hollow fibres onto grooved plastic strips was retained as the method of fibre
preparation to provide an orderly fibre pattern. Fibres were sandwiched between two grooved
plastic strips which were glued to secure fibre orientation as shown in Figure 4.1. Two sets of
glued strips acted to form an element which could be physically handled and built up into a
cartridge. Cartridges were built up into a module. Progression of construction is illustrated in
Figures 4.2 through 4.4.
The modules used for testing purposes are described in the following table. The fibre length of 5
cm was the same for all testing applications (Table 4.1).
13
-------
Table 4.1: Description of Modules and Fibres Used for Testing and Determining
Hydrodynamic Conditions Applicable for a Transverse Flow
Module
Use
Characteristics
Heaclloss analysis
Pervaporatioti testing to investigate
feed velocity effect
Pervaporation testing to investigate
the effect of temperature, permeate
pressure and organic type
Pilot testing to verify effect
of operating parameters
on pervaporation performance
4800 fibres
460nmOD
30 urn wall
890 fibres
540nmOD
125 urn wall
672 fibres
540nmOD
125 |im wall
5760 fibres
540 urn OD
125 nm wall
4.3 The System
A pervaporation system was required to evaluate these new modules and to quantify the effects of
certain operating parameters. This system required means of controlling feed flow rate, feed
temperature and vacuum pressure in order to test all operating parameters typical of pervaporation
operation.
Several options existed to provide methods for evaluating module performance. Two existing
systems were available to Zenon for testing module performance. The specifications of the two
existing systems are provided in Table 4.2. One system was a bench unit constructed by Zeton
Technologies Corporation for the Wastewater Technology Centre and the second system was a
pilot unit constructed by Zenon Environmental Inc. for the River Road Environmental Emergencies
Division of Environment Canada.
Since the bench unit provided better automation and control of system variables, it was used for
pervaporation testing. The unit was retrofitted to allow quantifying module performance with the
transverse flow module and for determining the effect of certain process parameters on overall
pervaporation performance. However, the pilot unit was also retrofitted to permit testing of a full-
scale module under typical pervaporation conditions that would expected in field operation.
14
-------
The bench pervaporation system was used to establish optimum hydrodynamics conditions for
pervaporation. A detailed description of the pervaporation system is provided in the Quality
Assurance Project Plan (QAPP) accepted by EPA (Appendix A) and will not be described further
in this report One correction to the QAPP, however, must be noted. The system volume as
measured by the addition of a tritium spike solution and measurement of the subsequent dilution
factor indicated that the system volume was in fact 10.3 L and not 6.0 L as stated in the QAPP.
Measurement of several tritium concentrations also indicated that complete mixing was achieved
within one minute. A schematic of the pervaporation bench scale system is provided in Figure
4.5.
A schematic of the pilot system is provided in Figure 4.6. This system was used to verify scale up
parameters. The membrane area for pilot operation is 100 times greater than the area used for the
bench testing work.
QUALIFICATION TESTING
The purpose of this section is to report the testing results for the transverse flow module and to
assess its potential for the removal of VOCs fern water.
Feed Pressure Loss
For estimating pressure drop through a bank of fibres, correlations for transverse flow are
available from the Engineering Sciences Data Unit (ESDU) series (1974). The pressure drop can
be estimated by
AP = 0.5KNiV2mix
where Nj is the number of rows in a module, Vmax is the velocity of the Liquid at greatest
constriction between the fibres and K is the friction factor coefficient and can be estimated from
K = a 0og(Re))2 + b (log(Re)) + c
where Re is the Reynolds number calculated using the outside fibre diameter, and the coefficients
a, b and c are dependent on the spacing between fibres and the spacing between rows. For the
current module design, the spacing (centre-to-centre fibre diameters) between fibres is 2.17 and K
may be estimated from the Table 5.1 by linear interpolation for each velocity. This estimate of K is
applicable for the range of Reynolds numbers from 10 to 1000.
15
-------
{
16
-------
5
£
17
-------
Liquid pressure drop tests were conducted with modules that contained approximately 4800 fibres.
An air/water manometer was used to measure pressure drop. A flow meter was used to monitor
the flow of water through the module.
The pressure drop for water flowing orthogonal through the array of fibres is shown in Figure
5.1. Tests were conducted at 20.7 °C, Results compare very well with the model (ESDU data) in
Figure 5.1 for such a configuration.
Table 5.1: Coefficients for Estimation of Pressure Drop in Transverse Flow of
Water Across a Bank of Fibres
Fibre Spacing
(diameters c/c)
2 0 -0.213 0.348
3 0 -0.161 0.013
18
-------
I
c
1000
800
600
400
200
100 200
Reynolds Number
300
5.2.
Figure 5.1 - Pressure Drop Through Array of 4800 Fibres
Pervanoration Results (Bench Scale)
5.2.1 Pervaporation Results
The results of the pervaporation tests are summarized in this section. All data included in this
section are used in a manner which satisfies the guidelines summarized in the QAPP regarding
data acceptance including whether data are used for verification or calculation purposes. In this
Section, the data represented in the figures is used for quantification only if the symbols or bars are
solid. Open symbols or bars indicate that data was only used for verification purposes.
The following major system variables were tested in the experimental program:
Feed flow velocity
operating temperature
Permeate side vacuum pressure
Type of organic compound
All run conditions and results axe summarized in Table 5.2. Runs 1,2 and 3 were conducted with
module 2. Subsequent use of module 2 was under conditions which damaged the fibres. A
different module (Module 3) was used for all other runs. The characteristics of Modules 2 and 3
are given in Section 4.
19
-------
I
I
a
1
I
E
£
3
I
I !
111
Hi
fills!! ii I If I ii
Ps§il! s » as § if
Is;SS12 s 5 ER I « «
iSiSUI ! ! I! i IS !!
HI
*< 4
HUH
3 ? ,1 3 ? ;
I!|ll III!!! Us
S9HI S9!iii sis
SI
33
-------
Results are best expressed as a separation factor and overall mass transfer coefficient. The
separation factor, a, is a measure of the effectiveness of separation and is estimated by
(YvOC/Y water)
a =
(XVOC/X water)
where Y is the concentration of the the indicated component in the permeate and X is the
concentration of the component in the liquid feed. Since a is dimensionless, X and Y may be any
convenient but consistent concentration unit. Although the separation factor is a convenient way of
communicating effectiveness of separation, it provides no useful information requited to design a
pervaporation system
The MTC (or k, with units of m/s) is calculated by
k = J/C
which defines the VOC removal rate and provides a method for calculating membrane area for a
pervaporation system J (expressed as kg/m2s) is the VOC flux and C (expressed as kg/m3) is the
concentration of the VOC in the water. The MTC allows flux data to be compared independently
of feed concentration.
The measured separation factor ranged from 2135 for EDC to a maximum of 41,000 for toluene.
This corresponds to mass transfer coefficient ranging from 1.71 x 10*5 to 11.1 x 10"^ m/s. The
significance of these results is given in the following section, grouped by important process
parameters.
The results of the pervaporation testing are included in Table 5.2. The reduced analytical and
system data is shown in Table 5.3. Data is included for both the module tested at increased
Reynolds Numbers as well as the module which was tested at other adjusted operating conditions.
Also included in these tables is the data for the pure water runs as it provides insight into system
operation. Runs marked with an asterisk (*) signify that some or all of the samples had to be
resubmitted for analysis due to lack of precision, In most cases this lack of precision was a direct
result of several aliquots measured as not 'useful' (as defined in the QAPP), which reduced the
sample space and therefore increased the variance beyond the precision requirements specified in
the QAPP.
5.2.2 Quality Assurance
This section documents the results of the Quality Assurance (QA) program required to validate the
findings of this study.
21
-------
Table 5.3: Reduced Analytical and System Data for Petvaporation Testing.
t««f
1
l»
1
t*
«
10*
M
11
12*
11
11
14
15
U
17
U
1*
a
u
a
a
M
a
a
17
a
a
a
)!•
11
a
n
M
a
a
«•
n
a
a
Moduk
Prawn
[Urr]
017
I JO
LM
0*4
1447
L17
1434
1145
4.11
14*
101
443
1.1*
L17
L04
LO)
444
15 JO
LM
LI5
LM
LA
44i
L14
13.74
LX
911
11.71
1J4
1531
1.45
L4»
1744
9J2
L44
MJO
174*
1431
105
)O
UO
MF
PT
Ate
r
SO.Mf
on
OM
OM
005
0X7
045
044
027
OM
010
005
an
004
005
OM
O07
00*
031
OM
007
007
003
OM
001
0.17
005
044
0.12
001
010
Oil
004
Oil
Oil
0 . 04
Oil
0 . 7
0.42
OM
RSD.MF
I1J*
50*
70*
IJ*
14*
54*
4.1*
1.7*
L4*
10*
10*
1J«
14*
11*
11*
17*
14*
14*
LI*
It*
14*
O7»
LOft
OM
LI*
L7*
13*
04*
OJ*
L2*
11*
L4«
U*
LI*
L4*
LO*
5 44*
14*
/It
PC]
144
14J
14.7
250
251
250
254
253
14J
344
14.9
250
247
150
144
150
154
144
J44
M0
04
174
173
»J
303
114
3)3
3)4
114
114
310
153
aj
354
34J
351
351
351
J53
SD.FT
Rso_nr r**4t
Ftowralc
SD.Ak
SD.MJ
F»W
C*M.
$o_r
RiltaUU
COCK.
SD.R
fUSCPM] i/L §/L
on
OM
OM
014
004
031
010
0.14
OM
045
007
014
O10
on
01*
on
033
on
045
on
041
04*
002
04*
Oil
OM
041
044
OM
on
OM
041
005
Oil
O17
OM
015
030
041
04*
01*
04*
04*
01*
11*
04*
10*
OJ*
14*
03*
O.T/o
04*
1.3%
04*
0>»
14*
OM
04*
OM
0.6%
14*
04*
OM
04*
Ol*
OJ*
0.5%
04*
0.3%
01*
11*
04*
OJ*
04*
0.1*
OJ*
034
11*
1.71
44)
114
1.43
L44
143
149
131
144
144
147
145
1.43
141
131
131
14*
Ul
14)
1.61
143
1.74
L74
La
La
1J4
Ul
U)
L77
1.70
LTD
131
137
134
L34
134
130
135
1.32
002*
0017
002*
0044
0044
0015
0015
0051
01*7
o0a
O147
0031
0.147
0011
0011
0.014
OM1
00tt
0075
0075
0074
0074
0045
0041
0041
0041
0041
0025
0025
0025
002)
002)
0021
002)
0021
000*
0004
0007
0004
0007
0007
000)
000)
0004
0001
000)
000)
0001
0004
0005
0005
0004
001*
0.018
004*
004*
0004
0004
0001
0004
0004
0004
0004
0.007
0007
0007
0.018
0011
0011
0005
0005
000*
00011
0003*
000)9
OOQ51
00047
00040
00341
00061
00061
00011
00011
00093
00014
00059
00061
00045
000)4
000)1
000)4
00051
0004*
00045
OOOM
00031
00034
O003t
o0oa
00047
00094
00101
00071
040*5
000*5
00105
00014
o.oou
OOM
0007
flft|4
0035
0014
0005
0001
0000
OW7
99000
0011
0015
99000
0010
0005
0417
0424
OOU
0011
0001
00M
0043
0004
0001
0005
OOU
OOM
OOU
0011
99000
Qffjj
0017
0044
00014
00014
00015
oocm
0004)
00011
00040
00044
00046
00044
00044
00041
00041
000)1
00011
000)1
00025
0001*
00015
00014
o0oa
00041
00011
00014
00022
00023
0002*
00051
00055
0000
00051
00050
0005)
00001
00010
0015
001)
99000
DOT)
OHM
0011
0005
0011
0004
0011
0025
0004
0014
0004
0014
0014
0010
000*
0007
OOU
0004
99000
0024
0011
0.010
001)
0037
oon
99000
0054
ooot
0.023
'trm«M4
C
0071
0044
0009
003*
0004
000)
0007
0007
0004
0.002
0.014
0004
00W
0070
0004
0075
0075
000*
0.031
Suotetf D*»tei»
R*I«K«S1
Potd Tarn
bfCV-C
tod ODBC
D
pmm
^»
-------
The results of the QA objectives for precision and completeness are shown in Table 5.4.
Completeness of test runs (approximately 85%) was lower than expected (90% expected) and
represents the fraction of the number of acceptable test runs divided by the total number of runs.
For a test run to be acceptable, the criteria outlined in the QAPP had to be satisfied. The two
criteria set in Table 2.1 of the QAPP included 1) Precision and 2) Accuracy. The precision results
for each test are included for each test in Table 5.3. Accuracy results were obtained by estimating
closure in mass balance and by submitting lab prepared standards in unmarked bottles along with
samples in order to get an unbiased estimate of the recovery error. 80% of the test runs closed the
mass balance to within the precision guidelines of the QAPP. A sample calculation for percent
recovery is shown in Appendix B. The recovery error for the program is reported in Table 5.4.
Both an average absolute and an average actual recovery error are provided to illustrate maximum
deviation. As the number of standards submitted (21 in total) for analysis increased, the average
actual recovery error tended towards zero and was not a good indication of equipment fluctuation
from test to test. Since the average actual recovery error does tend toward zero, it is an indication
that in the long term, the equipment and procedures were within the precision requirements set in
the QAPP.
23
-------
For the samples analyzed, 9 1% were within the precision requirements set out in the QAPP for the
standard deviation on aliquots submitted. Of all aliquots submitted, 82% of the samples had
internal calibration standards within the guidelines set out in the QAPP.
The overall number of tests was increased from 24 (estimated in the QAPP) to 39 in order to
compensate for the lower number of acceptable test runs as well as to provide some additional
water runs. Precision data was not available for feed flow-rate due to limitations in a transducer
used to convert a frequency signal to a variable current signal for the data acquisition system This
problem was only encountered at flow rates less than 11 L/min. (3 gpm). A totalizer on the
flowmeter was however tested for precision in the range used for the low Reynolds Numbers
pervaporation testing. The precision of the flowmeter was measured at 3.6% to 4.9% relative
standard deviation for flows less than 11 L/min. using calibration runs independent of the
pervaporation testing.
The QAPP was useful in identifying one problem area (correct estimation of system volume) in the
pervaporation test program. Use of system volume holdup, as determined by simple drainage of
the system, resulted in the inability to close a mass balance in the recovery of VOC's in the
pervaporation testing. Action was taken to determine the actual system volume holdup. To better
estimate system volume, a spike solution containing a measured quantity of tritium was added to
the system and the volume was estimated by the extent of tritium dilution in the final system
volume. Use of this system volume to estimate %R showed results that were within the objectives
of the QAPP.
No recalibration of analytical or system instrumentation was required throughout the sampling
program or the pervaporation testing. Daily analytical calibration checks were always within the
precision requirements set at the beginning of this test program. No test was rejected due to poor
precision or control of the operating parameters.
24
-------
Table 5.4: Quality Assurance Precision and Completeness Results
Completeness
Overall Number of Tests 85%
Samples Analyzed 91%
'Useful' Aliquots 82%
Percent Recovery for Mass Balance 80%
Recovery Error (avg. abs.) 0.026
Recovery Error (avg.) 0.003
Effect of Major Variables
Major process variables including feed flow velocity, operating temperature, vacuum pressure and
the type of compound are discussed in this section. These results were obtained using the bench
scale pervaporation system
Effect pf Reynolds Nymber
The effect of Reynolds number on the VOC removal rate (expressed as a mass transfer coefficient)
is illustrated in Figure 5.2. These results compare very well with the resistance in series model
(also shown in Figure 5.2) which was reported in earlier work (Lipski and C6t6,1990). These
results indicate that the models used for predicting mass transfer in transverse flow are in fact
applicable to closely packed fibres. The curve is characterized by a quick rise in mass transfer
followed by levelling off with increase in Reynolds number. The quick rise in mass transfer is a
direct result of the effectiveness of module design which promotes excellent removal of VOC's
from water even at low Reynolds numbers. Levelling off of the overall mass transfer coefficient
occurs because the membrane resistance becomes significant compared to the LFR at the higher
Reynolds numbers. To further increase VOC removal, at higher Reynolds numbers, membrane
thickness should be reduced or organic compound volatility increased (discussed further in this
section).
Replicate testing of the high Reynolds Number pervaporation testing was not possible because the
module was damaged at high velocity. The data were, however, adequate to confirm the models
for widely spaced hollow fibres, and were useful in predicting VOC removal performance for
closely spaced hollow fibres.
-------
5 .3 .2 E f f e c t Feed Temperature
The VOC permeability is one membrane factor affecting the separation (and hence removal) of
organics from water.
Increasing the feed temperature, increased the water flux through the silicone rubber membrane as
shown in Figure 5.3. Water flux data was obtained for both runs with pure water on the feed side
and water that contained VOCs. There was no measurable difference in these water fluxes. The
linear relationship of the water flux as a function of the inverse absolute temperature, is typically
found in pervaporation (as well as other physical systems which exhibit an Arrhenius-type
relationship), and is a measure of the membrane/liquid interaction. These water fluxes were
estimated both with and without organic contaminants, and showed no measurable difference in
the water flux. From this linear relationship, it is possible to predict water fluxes at higher
operating temperatures. The slope and intercept of this line were determined by linear regression.
120 -T-
100 . .
5
T 80 . .
U 60 . .
H 40 .
i
20 .
/
• D*.
— Model
SO 100 ISO 200 250 300
Reynold* Number
350
400
430
300
Figure 5.2 - Enhanced VOC Removal Promoted by Increase in Reynolds Number
26
-------
W*t«r Flu
16 -
14
12
10 -
8 -
1 .
a 0033
0.0034 a0033
21.1 29.9
T«jnp«r»t»r»
0.0032 (1>K ]
39.4 [«C
Figure 5.3 - Water Permeability Increased by Increasing Operating Temperature
This Arrhenius-type relationship for water flux was used in predicting operating performance in
Section 6.2.
The Arrhenius relationship is also applicable for the VOC/membrane interactions. As the
temperature increases, the VOC permeability through the membrane increases, just as for water.
This increases the potential of the membrane to remove VOCs from water. Increased permeability
could not however be shown experimentally at the low Reynolds Numbers due to liquid film
resistance (LFR) that controls the rate of VOC removal. To illustrate, Figure 5.4 shows that the
removal of VOC at 25°C and 35°C arc not significantly different at low Reynolds number for either
TCE or toluene, due to LFR. Toluene removal increased from approximately 50 jim/s to 60 jinVs
while it dropped for TCE from 50 [ua/s to 48 jinVs. These differences are not outside the limits of
the precision of the study and do not represent any significant change in the rate of VOC removal.
Although demonstration of higherpermeabilities at higher temperatures, shouldbc expected at
higher Reynolds Numbers, poor fibre stability at higher temperature and Reynolds numbers did
not allow testing under these conditions.
Higher operating temperatures have an adverse affect on the separation factor for systems that are
liquid film controlled. Increases in temperature cause increased water permeability with no
increase in removal of VOCs. The increased water fraction in the permeate actually dilutes the
27
-------
permeate, resulting in a net decrease in the separation factor as illustrated in Figure 5.5. For VOCs
such as EDC that are not liquid film controlled, increasing feed temperatures also reduces the
separation factor, but not to the same extent as VOCs that are LFR controlled.
In summary, it may be generalized that operation at higher temperatures in a pervaporation system
that is liquid film controlled does not increase the removal of the VOC for high vacuum operation.
It is therefore necessary to adjust other conditions such as increasing Reynolds number or
increasing the membrane thickness in order to regain the high separation factors while at the same
time derive the benefits from operation at higher temperatures (see Section 4.25).
5.3.3 Fffect of Different. M"Tfc
As indicated earlier, the removal of VOCs through a membrane is dependent on the preferential
partitioning of the VOCs out of the water and in this case onto the membrane surface. Some
organic compounds are more volatile and partition more readily than other compounds.
Differences in Henry's Law constant rather than permeability of VOCs through the membrane
itself (see Section 24) are better used to describe the effectiveness of removal. As such, it was
important to quantify the rate of removal as a function of volatility. Three compounds tested are
provided in Table 5.5 (Montgomery and Welkom, 1990). TCE doubles in volatility from 25°C to
37°C and at 25°C, TCE is approximately 30% more volatile than toluene. TCE is 10 times more
volatile than EDC at 25°C. The results in Figure 5.4 indicate that the overall mass transfer
coefficients (at high vacuum) for TCE (at 25°C and 35°Q and toluene are indistinguishable. In
addition,TCE and toluene removal is only three to four times higher than the removal rate for EDC,
although EDC is only one-tenth as volatile as TCE. This is in agreement with the resistance-in-
series model which generalizes that for very volatile compounds, the liquid film boundary layer
becomes rate controlling and that rate of removal is independent of the membrane and VOC
properties. On the other hand., the less volatile EDC is not partitioned readily to the membrane
surface and, therefore, LFR does not play a critical role in defining VOC removal. Hence,
removal of EDC from water was increased by increasing temperature from 25°C to 35°C as shown
in Figure 5.4. Testing at temperatures above 35°C with the transverse flow module was not
possible with the current support fibre without collapsing the fibre.
Just as increases in the permeability (illustrated in Section 4.2.2) had no effect on organic removal,
enhanced partitioning at the higher temperatures does not aid in removal of the VOC in a system
that is liquid film controlled. For TCE and toluene, removal of the VOC from the bulk liquid is
governed by module hydrodynamics. Even if Henry's Law Constant for TCE is doubled, by
28
-------
increasing the feed temperature to 35°C (from 25°Q the TCE removal is not measurably changed
because removal is limited by LFR.
Table 5.5; Organic Compounds Considered for Removal by
Pervaporation at Various Operating Temperatures
Compound
Temperature
ra
Henry's Law
Constant
[atm-m3/mol]
Toluene
TCE
25
25
37
25
0,0067
0,0091
0.0196
0,00091
Coefficient
60 •
5 0 •
40 *
30 .
20 •
10 .
0 .
-_4-
II | I | 1 1-t.
I....1 J l, ,
1
,l.li™l, J,
I £ • t • *
r
« 11 II"" 1 1 T"! 1 1 1 "I 1 1 TTTTTTT1 — >"f"t
SM 66 J B B 1 is
-23«C-
->k-33sC->BI°Ck 35°C-
Figure 5.4 - VOC Removal for Toluene, TCE and EDC @ 25 Through 35°C Using
a Transverse Flow Pervaporation Module Operating Under High
Vacuum and a Water Side Reynolds Number of 40 to 60
29
-------
23000
20000 .
15000 -
S«p*r*tloB
Factor
10000 •
3000
o P i p i i i i i i i n M i I'M < n i i i i i M i i»i 11 i»i i M i i i I'l
II I 8 E I B
-25*C-
>k-33*C->01eCk 35°C
Figure 5.5 - Separation Factors for VOC Removal for Toluene , TCE and EDC @
25 - 35°C Using a Transverse Flow Pervaporation
Module Operating Under High Vacuum and a Water
Side Reynolds Number of 40 to 60
5.3.4 Permeate Pressure
Vacuum pressures of 1000 Pa or greater are typically employed in pervaporation applications.
Most vacuum pumps can however provide vacuum pressures much less than 1 Pa. Operating
vacuum pressures greater than 500 Pa are termed 'rough vacuum' in the vacuum pump industry.
The ultimate vacuum achieved in typical pervaporation applications is not Controlled by the vacuum
pump, but rather by the condenser upstream of the vacuum pump (Figure 2.2). The temperature in
the condenser determines the system vacuum pressure (as governed by vapour pressure of the
permeate components trapped in the condenser). The reduction in flux through the membrane
caused by system operation away from the ultimate vacuum is considered a vapour side restriction,
(VSR), in the pervaporation process.
In order to simulate an industrial application of pervaporation, permeate flow was constricted to
obtain a rough vacuum in the range from 100 Pa to approximately two-thirds of the saturation
pressure (calculated at the feed temperature) of the water. The effect of the permeate pressure on
the water flux is shown in Figure 5.6. The vapour pressure fraction is estimated from the total
permeate pressure, P, divided by the vapour presure, Pvan» of water at the feed temperature. The
reduction in water flux is defined as 100% reduction when the flux is zero and 0% redcution at
30
-------
high vacuum where flux is maximized for a given temperature. The linear model suggested by
Lipski eliL (1991) estimates, quite reasonably, the reduction in flux due to VSR's. The data was
obtained for operating temperatures in the range of 18°C up to 35°C for pure water and water that
contained up to 10 ppm organic compounds. There was no measurable difference in water flux in
the presence of VOCs.
The reduction in the organic flux due to VSR is shown in Figure 5.7. Similar to Figure 5.6, this
organic flux is linearly dependent on the vapour pressure of the water (at the membrane) rather
than the vapour pressure of the organic compound. Since the organic compound is the minor
component in the permeate, the water vapour acts to sweep away the organic from the membrane
surface and thus reduce the partial pressure of the VOC, effectively by dilution. If the water
vapour is considered as a plug of material that may contain some fixed quantity of VOC, then
increasing the flow of that plug increases the removal of VOCs from the membrane and the water
(It was estimated that the permeate vapour was saturated with the VOC in all cases where removal
of VOCs from water was observed and VSR had an effect). In effect, the quicker the water can be
removed from the membrane surface, the faster the organic can be swept away from the membrane
surface and greater removal of the VOC from water can be achieved. On the other hand, if the
vapour pressure at the membrane approaches the downstream pressure, the plug of water vapour
becomes stagnant and VCK! removal effectively stops since the VOC must diffuse through the
water vapour to leave the membrane surface. This difiusion velocity is very small compared to the
sweep (bulk flow) velocity which is generated if the water vapour can be removed from the
membrane surface. In most cases, however, sweep of the water vapour and reduction of the VOC
partial pressure is maintained to provide continuous removal of VOCs. Furthermore,
pervaporation is not limited to removal of VOCs at ppm levels. Since the water vapour acts to
dilute the VOC in the permeate, and so long as the water is being swept away from the membrane
surface, removal of ppb levels of VOCs is possible (C6t£ and Lipski, 1991).
535 Effect of Feed Temperatures for Roug'hVaccum Operation
As stated above, the permeate side pressure is a function of the condenser temperature. Since ice
buildup in condensers poses a handling problem, condenser temperatures should be maintained
above freezing. Typical condenser pressures as a function of condenser temperature are illustrated
in Figure 5.8. Choosing a condenser temperature of say 5°C, the condenser pressure (estimated
from vapour pressure data available for the organic and aqueous phase) is estimated at
approximately 16 torr. Given a fixed permeate pressure (2,000-2,500 Pa), increased VOC
removal, shown in Figure 5.9, was demonstrated in this study by increasing the feed temperature.
Removal was increased by 3 to 4 times when Ac feed temperature was increased from 25°C to
-------
1,2 -r
PUx
0.1 . .
0,6 - •
O6 0.7 0.8 0.9
0,4 . .
0,2 . .
Figure 5.6 - Reduction in Water Flux with Increase in Module Permeate Pressure
0 9.1 0.2
0.3 6,4 CJ 0.6 6.?
Fmctloa of W»l«r V«por Pre»wr«
0.8 0.9 I
Figure 5.7 - Reduction in VOC Flux with Increase in Module Permeate Pressure
-------
30 T
Coxknw Prtuure
(tocr)
9 l§ II 12 13 14 IS
Coodowor T«nperttw« |*Cj
Figure 5.8 - Condenser Pressure Estimated as a Function of Condenser
Temperature for a Saturated Toluene/Water Liquid
13 -r
30 . •
Man Truttftr
c»ctn«tt»t
ti.-/.] a I
10 ..
s..
10
4.
26 31 JO
Tetnp*r«tDf* ("C]
Figure 5.9 - Increased VOC Removal at High Temperature for Rough Vacuum
Operation
33
-------
35°C for simulated rough vacuum operating conditions. This increased removal, in rough vacuum
operation, is directly attributed to reducing the VSR for water at the higher feed temperature. As
the VSR is minimized, removal rates will increase until rates are comparable to rates achievable
under high vacuum operation.
It is apparent that for rough vacuum, the effective rate of removal of a VOC from water may be
increased by increasing the feed temperature. Increasing the feed temperature (which increase the
vapour pressure at the feed), increases the driving force from the feed to the condenser (the
condenser being at a constant temperature and pressure). Extrapolating, VOC removal can be
improved to within 5% of the maximum removal defined by the LFR and the membrane resistance
by operating at feed temperatures of 75°C. The 5% shortfall in maximum VOC removal is due to
VSR. Although operation at higher temperature does not increase removal rates due to higher
partition coefficients or membrane permeabilities removal is enhanced for systems by reducing the
VSR
Improved performance can, however, be realized for EDC and other semi-volatiles by operation at
higher feed temperatures. As feed temperatures increase and volatility increases (i.e. Henry's Law
Constant for EDC more than doubles from room temperature up to 37°C, 0.00225 atm-mtymol),
greater removal can be achieved for these compounds which were not liquid film controlled at
the lower operating temperatures. As with removal for the volatile compounds, removal rates can
only be improved for semi-volatiles to a rate where LFR becomes rate controlling. At such a
temperature, removal rates of semi-volatiles will be comparable to removal rates for volatiles.
34
-------
5.4 Pilot Scale Operation
The River Road pervaporation pilot unit was tested on a synthetic wastewater. The synthetic
wastewater contained approximately 3 ppm toluene in tap water. The system was operated under
conditions that allowed accurate monitoring of process parameters. Flowrates were adjusted to
provide 1) reasonable toluene removal, and 2) effluent concentrations well above the method
detection limit for toluene analysis. Accurate analysis of toluene concentration was critical in
assessing the actual rate of removal by pervaporation.
One pervaporation test was conducted using a module that contained 5760 fibres providing a
surface area of 0.5 m2. Operating conditions for this test are provided in Table 5.6. Feed and
retentate samples were taken at the beginning, middle and end of the run to compare to expected
pervaporation performance. The model (for estimating removal rate as a function of Reynolds
Number and other process parameters) which was verified in Section 5.3 was again used to
estimate toluene removal now using a module with ten times greater surface area. The measured
toluene removal was in very good agreement with the model and was well within the analytical
precision limits. This test indicated that models used in Section 5.3 are accurate for scale-up
purposes. Although removal of more than 90% could be achieved by increasing the membrane
surface area or reducing the volumetric flowrate, such removal could not be verified analytically,
and, no such tests were performed.
Verification of these models confirms that the technico-economical analysis (Lipski and C6t6,
1990) used for estimating effectiveness of removal can be used for scale-up purposes.
35
-------
Table 5.6: Operating and System Parameters for Pilot Testing
of a Pervaporation Module
Membrane AreaCni2) 0.5
Operating Feed Temperature (°Q 35
Condenser Temperature (°Q -10
Condenser Pressure (ton) 3.5
Influent Flowrate (litres/min.) 2.2
Reynolds Number 20
Removal Rate (%)
At start of test 42
In middle of test 41
Endoftest 47
Removal predicted by Model (%) 42
6 . 0 ppnrpss OPTIMIZATION
The purpose of this section is to outline design and operating conditions that will maximize
pervaporation performance. The basic model used to perform the optimization is illustrated by
Lipski and C6t6 (1990). The results will only be summarized here.
6.1 Operating
To increase removal rates, higher Reynolds numbers should be developed to overcome liquid film
resistance (LFR) and higher feed temperatures should be utilized to reduce the vapour side
restrictions (VSR). In addition, higher operating feed temperatures will allow removal of semi-
volatile compounds from water. For operation at increased Reynolds number or operating
temperature the substrate fibre material must be strengthened. In addition, permeate pressure
should be minimized to reduce VSR A case study considering these system conditions will be
illustrated in Section 6.2
6 . 2 Case Study
The benefits of employing the changes in system's operation which were recommended in Section
6.1 are compared to the current capability of the existing pervaporation module. The significant
-------
limitation of the existing module is that operation is limited to low Reynolds Numbers and feed
temperatures. A computer model was used to provide capital and operating requirements for the
available system (used in this study) and compare them with the requirements of an improved
system The current system is defined as having a ceiling temperature of 35°C and a maximum
Reynolds Number of 70, while an optimized module and system would have a ceiling temperature
of 75°C and be capable of achieving a Reynolds Number of 600. To provide an equivalent basis
for comparison, capital equipment is considered that will provide and support 99% removal of
toluene from a water stream. For purposes of comparison, support equipment (i.e. condensers,
feed and vacuum pumps) for both systems were identical. Membrane thickness was increased in
the high temperature application that would provide an equivalent water flux in comparison to the
low temperature constraint and thus call for an identical vacuum pump and condensation train.
Pervaporation operation was considered at the high feed side Reynolds number represented in
Figure 5.2 and operating at a feed temperature of 75°C extrapolated from Figure 5.3 and using
available Henry's Law data. The results of such a case study is provided in Table 6.1. The model
considered two systems, each to provide 99% removal of toluene from a 34 litre/min. water
stream. The condenser temperature was 5°C which would provide a vacuum pressure of 16 torr.
The significant component in the capital cost is the membrane cost. The reduced membrane
requirement at the higher Reynolds Numbers and feed temperature, directly impacts on the capital
cost of the pervaporation system. Since a significant component of the operating cost includes
capital depreciation, reducing capital cost reduces the operating cost directly. Energy requirements
for both systems are however equivalent and are approximately 0.17 kWh/nA These costs are in
line with previous cost schedules for pervaporation (Lipski and C6t6,1990).
6.1: Benefits of Operation of Pervaporation System at Optimum Conditions
Feed Temperature (°Q
Reynolds Number
Mass Transfer Coefficient (nm/s)
Flux Reduction (VSR, %)
Membrane Requirement (m2)
Capital Cost ($K)
operating Cost (Vm3)
35
52
43
38
98
217
4.35
75
633
268
6
10
31
0.57
37
-------
6.3 Technical and Economic Analysis
To carry out an economic analysis, a case study considered a 167 litre/min. system which
contained 10 ppm trichloroethylene (TCE) that was to be reduced by 99%. Pervaporation was
compared with existing technologies (including assumptions) such as air stripping and activated
carbon (Lipski and CM, 1990). The estimated cost of treatment by air stripping alone was
estimated at $0.10/m3. The cost of combining air stripping with granular activated carbon (GAC)
in various configurations, varied from $ 0.40/m3 for air stripping and vapour-phase GAC, to $
0.80/m3 for liquid phase GAC alone. These costs include regeneration or disposal of activated
•j
carbon. By comparison, the cost of pervaporation would be in the order of $ 0.56/nr using the
membrane and module design developed in this project. This cost was estimated from energy
requirements and amortization of assembly costs and component costs and indicates that
petvaporation is cost competitive with existing technologies.
Incineration costs are not included in the above costs. Since it is necessary to incinerate only the
organic phase of the pervaporation condensate or from the carbon regeneration, incineration costs
would represent a fractional increase in the overall operating cost. If incineration costs of up to
$1/liter (depending on fuel value) are assumed treatment costs would increase by only $0.10 /m3
if effluent concentrations are up to 100 ppm and only the organic phase is sent off for incineration.
For industrial applications where the effluent contains a single organic there exists the potential for
organic reuse and these incineration costs are not applicable.
To obtain and optimize treatment costs for pervaporation a computer cost model package was
developed (Lipski and C6t6, 1990). A sensitivity analysis was carried out to identify critical
variables and to optimize these variables.
Technically, pervaporation has several advantages over carbon as described below. Since air
stripping alone is not suitable for groundwater remediation and requires carbon adsorption for
controlling off-gas emissions, air stripping alone can not be compared to pervaporation. It must
however be mentioned that pretreatment is often required to avoid precipitation and fouling in air
stripping columns. Some of the major advantages that pervaporation can offer over carbon
adsorption are:
1) Pervaporation uses no sorbents which must be regenerated
2) Continuous monitoring is not necessary for effluent breakthrough.
38
-------
3) Pervaporation is continuous and offers immediate recovery of solvents for
industrial applications.
4) Pervaporation is suitable for both high and low concentration VO contaminated
water.
5) Pervaporation offers removal of moderately soluble compounds such as
ethylene dichloride which can not be removed by carbon.
Carbon adsorption is however applicable in certain instances. Use of activated carbon is best
suited for water contaminated at low VOC concentrations as monitoring for breakthrough and
column saturation will be infrequent. Used downstream of pervaporation, carbon will be effective
for removal of any residuals not suitable for pervaporation as well as removal of any traces of
VOCs without significant loading on a carbon bed.
7.0 CONCLUSIONS AND RECOMMENDATIONS
The work in this project has developed and identified improved pervaporation operation for VOC
removal as compared to use of pervaporation modules existing on the market prior to this study.
This improvement was measured as a reduced membrane requirement for any given application
and a more energy efficient pervaporation system. Secondary objectives which were accomplished
in order to deliver these results are included below.
Hollow fibres were developed with a thick and selective layer on the outside of a hollow fibre.
This active silicone rubber layer was thick and continuous from the inside to the outside of the
hollow fibre and used a microporous membrane as a support to facilitate membrane preparation.
This membrane thickness (125 \un) was optimized to provide a strong and selective active layer.
A prototype transverse flow module was developed in this work. This module consisted of
hollow fibres (540 Jim OD) spaced 1 mm centre to centre in both lateral and longitudinal direction.
These modules were used to test pervaporation performance. Throughout the testing program,
there was no evidence of fouling or channelling of feed in the transverse flow modules, Removal
rates were shown to increase beyond the rates reported in previous work. Mass transfer and
pressure drop correlations available for widely spaced tubes were validated for closely packed
hollow fibres. Hydrodynamic conditions were optimized to provide long membrane and module
life and good VCK! removal. Although VOC removal could be enhanced under very turbulent
conditions, such operation reduced fibre life. Bench testing also indicated that rate of removal is
independent of temperature and type of VOC for the more volatile compounds (such as TCE and
-------
toluene) for low Reynolds Numbers and under conditions of very high vacuum. Removal was
more effective with increase in Reynolds Number or by increasing feed temperatures for rough
vacuum operations. Limited testing at pilot scale confirmed the bench-scale results.
Models and equations describing hydrodynamic conditions, developed for other applications, were
confirmed in this work and could be used as a tool for estimating performance under all
hydrodynamic conditions. A computer model was developed using these equations and enabled
identification of key operating parameters for pervaporation operation. Optimization of key system
variables led to identification of process conditions for improving pervaporation performance:
• Semi-volatile compounds such as EDC and methylene chloride can be removed more
effectively by pervaporation at higher feed temperatures.
. To improve overall performance of a transverse flow pervaporation system, a module should
be developed for operating at a Reynolds Number above 600 and at a temperature of at least
75°C.
These module improvements should be implemented prior to any field testing.
Field testing of pervaporation will be required before commercialization of pervaporation can be
exploited as an alternative site or industrial remediation process for VOC removal. Since
peavaporation has distinct advantages over other technologies when considering high concentration
VOCs, initial demonstration and field testing should focus on high concentration effluents. In
particular, single component VOCs which have some reuse value should be considered in order to
demonstrate the VOC recovery potential of pervaporation.
Commercialization of pervaporation will require scale-up of pervaporation modules so that
membrane costs may be reduced Other areas for improvement in pervaporation systems will be
identified at pilot scale.
40
-------
References
Brim, J.P., "Etude thermodynamique du transfert sdlectif par pervaporation k travers des
membranes 61astomeres d'especes organiques clissoutes en milieu aqueux." These de Doctoral
(TEtat, Paris (198 1).
C6te", P.L. and Lipski, C., "Mass Transfer Limitations in Pervaporation for Water and
Wastewater Treatment", Proceedings of the Third International Conference on Pervaporation
Processes in the Chemical Industry, Nancy, France, September 19-22.1988.
C6t6, P.L. and Lipski, C., "The Potential of Pervaporation for Water Treatment", Proceedings
of the Fifth Annual HAZTECH Canada Conference, Toronto, May 14-15,1991.
Engineering Sciences Data Unit. Pressure Loss During Cross-Flow of Fluids with Heat
Transfer over Plain Tubes without Baffles. 1974, Item No. 74040, Engineering Sciences Data
Unit, London.
Eustache H. and G. Histi, "Separation of Aqueous Organic Mixtures by Pervaporation and
Analysis by Mass Spectrometry or a Coupled Gas Chromatograph - Mass Spectrometer",
Journal of Membrane Science. 8(1981)105.
Federal Register, Volume 52#74, page 12866-12874, April 17.1987.
Niihuis, H., "Removal of trace organics from water by pervaporation; a technical and economic
analysis", Ph.D Thesis, University of Twente, 1990.
L6veque,J.A.,AmmMines, 13,201,305,381 (1928).
Lipski, C., P.L. C6t6, and H.L. Fleming "Transverse Feed Flow for Hollow Fibres
Significantly Improves Mass Transfer at Low Energy Consumption", Proceedings of the Fii
International Conference on Pervaporation Processes in the Chemical Industry, Heidelberg,
Germany, March 12-15, 1991.
Lipski, C. and C6te\ P.L., 'The Use of Pervaporation for the Removal of Organic
Contaminants from Water", Environmental Progress. 9(^:254-261.1990.
41
-------
Montgomery, John H. and Linda M.Welkom, Groundwater Chemicals Desk Reference, Lewis
Publishers Inc., Michigan, 1990.
Nguyen, T.Q. and K. Nobe, "Extraction of Organic Contaminants in Aqueous Solutions By
Pervaporation", Journal pf^fcpifaaqg-Scisncg.SQd 987^1 1 .
Psaume, R., "Application de la pervaporation au traitement de 1'eau potable: elimination de
d6riv6s halog6n6s & l'6tat de traces", thfcse de doctorat, Institut National des Sciences
Appliqu&s de Toulouse, 1986.
Radian Corporation, "Air Stripping of Contaminated Water Sources - Air Emissions and
Control", EPA-450/3-87-017, 1987.
Radnoff, D. and C. Lipski, "Laboratory Scale Pervaporation Testing for Spills and Leachates",
Proceedings of the ffl TechnicaLSeminar on Chetnjgal SpJUs? February 8-11,1988.
Shen, T.T. and Sewell, G.H., "Control of VOC Emissions from Waste Management
Facilities". Journal of Environmental Engineering. 114(1988)1392.
Yang, M.C. and Cussler, E.L., 'Designing Hollow Fibre Contactors", AIChEJournal.
32(1986)1910.
42
------- |