EVALUATION OF ULTRAFILTRATION  TO RECOVER
     AQUEOUS  IRON PHOSPHATING/DEGREASING  BATH
                        by
Gary D. Miller, Timothy C. Lindsey, Alisa  G.  Ocker
  Hazardous Waste Research and Information Center
Illinois Department of Energy and Natural Resources
            Champaign, Illinois   61820
                Michelle C. Miller
       Daily and Associates Engineers,  Inc.
            Champaign,  Illinois   61820
              Contract No. CR-815829
                  Project Officer

                   Paul  Randall
       Pollution Prevention Research Branch
       Risk Reduction Engineering Laboratory
              Cincinnati, Ohio 45268
       RISK REDUCTION ENGINEERING LABORATORY
        OFFICE OF RESEARCH AND DEVELOPMENT
       U.S. ENVIRONMENTAL PROTECTION AGENCY
              CINCINNATI,  OHIO  45268

-------

-------
                                         NOTICE

       The information in this document has been funded wholly or in part by the United States
Environmental Protection Agency (U.S. EPA) under Contract No. CR-815829 to the University of Illinois
and the Hazardous  Waste Research and  Information  Center in Champaign, Illinois.  It has  been
subjected to the Agency's peer and administrative review process,  and it has been approved  for
publication as an EPA document.  Approval does not signify that the contents necessarily reflect the
views and policies of the  EPA, nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.

-------
                                         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 the EPA to perform research to define our environmental problems,  measure
the impacts,  and search for solutions.

       The 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, Superfund-
related activities, and pollution prevention. This publication is one of the products of that research and
provides a vital communication link between the researcher and the user community.

       This report describes the results of laboratory and field testing of an ultrafiltration system to
recover and reuse  an iron phosphating/degreasing bath. This pollution prevention  project supports the
emphasis on reducing  generation of hazardous waste by encouraging study and development of
methods to effectively use and recover aqueous cleaning  solutions.
                                               HI

-------
                                         ABSTRACT

       Pollution prevention efforts targeted the hazardous waste generated from a 5000-gallon iron
phosphating/degreasing bath  used by a meta! fabricator to clean and precondition  steel parts for
painting.   With extended use, the  buildup  of emulsified oil  in  the bath reduced cleaning  and
phosphating efficiency. Dragout of oil from the bath into the rinse water also pushed oil and grease
levels in the effluent over the  allowable limit.  When oil in the bath began to sacrifice  product quality
and effluent levels edged closer to the maximum allowable limit, all 5000 gallons were dumped and
replaced. Periodic dumping, about three times each year, resulted in at least 15,000 gallons per year
of hazardous waste. Several waste minimization alternatives were considered, and ultrafiltration was
selected as the most promising technology to recover and reuse the bath and to reduce the total
amount of  hazardous waste  generated.  Ultrafiltration  has proven successful in  similar industrial
applications with alkaline cleaning solutions, but the application of new membrane filtration technology
to this acidic, corrosive, high temperature bath was an innovative approach to pollution prevention.

       This project was carried out in four stages:  (1) initial assessment of the problem and evaluation
of alternatives, (2) bench-scale screening of ultrafiltration membrane candidates, (3) pilot-scale study
at the Illinois Hazardous Waste  Research and  Information  Center (HWRIC), and (4) full-scale
implementation and testing onsite at the company's facility.   Full-scale testing integrated the new
waste reduction scheme into the facility's production process by applying ultrafiltration directly to the
5000-gallon iron phosphating/degreasing bath. Ultrafiltration successfully removed oil contamination
from the bath and returned clean process solution back to the original 5000-gallon tank. Ultrafiltration
concentrated the hazardous component down to  10 gallons of oily waste and reduced hazardous waste
generation 99.8%. The concentration of oil in the bath was substantially reduced and maintained at
acceptable  operating levels.  Permeate flux  rates exhibited excellent performance  and  were high
enough to compete with the constant input of oil from the production line. A significant portion of the
unused phosphating agents were also conserved although some surfactant was lost.  Product quality
tests revealed that quality achieved during the full-scale ultrafiltration study was good for the facility's
application.  The estimated payback period associated with implementing ultrafiltration was only 6.9
months.  Results  of this study were used to justify installing a permanent ultrafiltration system and
operating practices that would improve product quality.

       This report  was  submitted in partial fulfillment  of  Contract No. CR-815829  under the
sponsorship of the U.S. Environmental Protection Agency. This report covers a period from February
1992 to November 1992 and work was completed as of November 1992.
                                              IV

-------
                                   TABLE OF CONTENTS
Foreward	iii
Abstract	iv
Figures	vi
Tables  	vii
Abbreviations and Symbols	  viii
Acknowledgements	ix

       1. Introduction	•	  1
              Project Goals and Objectives  	  1
              Industrial Participant	  1
              Problem Description  	  3

       2. Process Background  	  6

       3. Waste Reduction in the Metal Fabricated Products Industry
              Waste Generation	  9
              Aqueous Cleaners	   10
              Ultrafiltration	   13

       4. Bench-Scale Membrane Screening
              Methods and Materials	   19
              Bench-Scale Activities	   19
              Results and  Discussion	   21

       5. Pilot-Scale Testing
              Methods and Materials	   23
              Pilot-Scale Activities	   25
              Results and  Discussion  	   26

       6. Full-Scale Implementation and Testing
              Methods and Materials  	   31
              Full-Scale Activities	   31
              Results and  Discussion  	   32

       7. Quality Assurance	   45

       8. Economic Analysis	   50

       9. Conclusions and Recommendations	   53

 References	   55
 Glossary	   57
 Appendix  	   58

-------
                                        FIGURES


Number
                                                                                   Page
1             Floor Plan of Cleaning and Painting Operations	  2
2             Flow Diagram of Main Operations and Associated Waste Stream  	  4
3a            Tubular Membrane Module	.16
3b            Spiral Wound Membrane Module  	  1 g
3c            Hollow Fiber Membrane Module  	             1 Q
4             Bench-Scale Ultrafiltration Apparatus	'.'.'.'.'.'.'.'.'.  20
5a            Comparison of Wastewater Flux Rates During Bench-Scale Testing   	  22
5b            Comparison of Normalized Flux Rates During Bench-Scale Testing	  22
6             Continuous Scheme Ultrafiltration	             24
7             Batch Scheme Ultrafiltration	!!!!.'!  24
8             Modified-Batch Scheme Ultrafiltration	'.'.'.'.  24
9             TOC in Process Tank and Permeate vs. Time
                    Using Clean Process Solution	  27
10            Phosphate in Process Tank and Permeate vs. Time
                    Using Clean Process Solution	  27
11 a           TOC in Process Tank and Permeate vs. Time
                    Using Waste Process Solution	  28
11 b           Temperature Effect on Permeate TOC vs. Time
                    Using Waste Process Solution	  28
12            Phosphate in Process Tank and Permeate vs. Time
                    Using Waste Process Solution	  29
13            Calibration Curve for Dura-Gard TOC	['.'.'.  33
14a           Oil and Surfactant TOC  in Bath vs. Time  	' '    37
14b           Oil and Surfactant TOC  in Rinse vs. Time	' ' '  37
14c           Oil and Surfactant TOC  in Process Tank vs. Time	  38
14d           Oil and Surfactant TOC  in Permeate vs. Time	'.'.'.'.'.  38
15            Phosphate in Bath, Process Tank, and Permeate vs. Time	  40
16a           Permeate Flux Rate vs. Time  	\\  42
16b           Permeate Flux Rate vs. Time (Extended Study)	  42
                                           VI

-------
                                        TABLES
Number

1            Typical Materials Used and Types of Wastes
                    Generated in the Metal Fabricated Products Industry ..............  10
2            Membrane Candidates and Material Specifications .....................  19
3            Product Quality Test Results  ...................................  41
4            Sample Summary ........................................ ...  45
5            Precision Data for Full-Scale Sample Analysis  ........................  48
6            Accuracy Data for Full-Scale Sample Analysis ........................  49
7            Assumptions for Economic Analysis ............................. .  50
8            Cash Flow Summary  ........ . .................................  51
9            Economic Summary  ................................... ......  52
                                           VII

-------
ABBREVIATIONS

ASTM
cmc
COD
DAF
EDTA
EPA
HSWA
HWRIC
1C
IR
IRR
kW
L
MEK
mg
MIBK
mL
mm
MWCO
N
NPV
PHC
POTW
PPA
psig
PVDF
QAPP
RCRA
RPD
SARA
TCLP
TOC
TRI
UF
VOC
WRITE

SYMBOLS

Ca
CaCO3
Cr03
Fe2(P04)3
Mg
NaCI
NaOH
Na2SO4
PH
PK.
P04
                              ABBREVIATIONS AND SYMBOLS
 American Society for Testing and Materials
 critical micelle concentration
 Chemical Oxygen Demand
 Dissolved Air Flotation
 Ethylenediaminetetraacetic acid + its sodium salts
 Environmental Protection Agency
 Hazardous and Solid Waste Amendments to RCRA
 Hazardous Waste Research and Information Center
 Ion Chromatograph
 Infrared Spectrophotometer
 Interest Rate of Return
 kilowatt
 Liter
 Methyl Ethyl Ketone
• milligrams
 Methyl Isobutyl Ketone
• milliliter
• millimeter
• Molecular Weight Cutoff
• Normal
• Net Present Value
• Polyhydrocarbon
• Publicly Owned Treatment Works
• Pollution  Prevention Act
• pounds per square inch, gage
- Polyvinylidene di fluoride
• Quality Assurance Project Plan
• Resource Conservation and Recovery Act
- Relative Percent Difference
- Superfund Amendments and Reauthorization Act .
- Toxicity Characteristic Leaching Procedure
- Total Organic Carbon
• Toxics Release Inventory
- Ultrafiltration
- Volatile Organic Compound
- Waste Reduction Innovative Technology Evaluation
 Calcium
 Calcium Carbonate
 Chromic Acid
 Iron Phosphate
 Magnesium
 Sodium Chloride *
 Sodium Hydroxide
 Sodium Sulfate
 Log of Hydrogen ion concentration
 Dissociation constant
 Phosphate
                             viii

-------
                                   ACKN 0 WLEDG EM ENTS

       The U.S. Environmental Protection Agency and the Hazardous Waste Research and Information
Center acknowledge the important contributions made by the Civil  Engineering Department at the
University of Illinois,  R.B.  White, Inc., Koch Membrane  Systems, Inc., Arbortech Corporation,
Osmonics, Inc., DuBois Chemicals, Inc., and John Sparks of the U.S. EPA Global Change Division. The
support and cooperation of each  of these people  and organizations was greatly beneficial to this
project.
                                              IX

-------
                                           Section 1
                                       INTRODUCTION

       This project is one of five projects conducted under the State of Illinois - U.S. Environmental
Protection Agency's (EPA) Waste Reduction  Innovative Technology Evaluation (WRITE) Program.  It
represents a joint research effort of R.B. White, Inc., Bloomington, Illinois; the Illinois Hazardous Waste
Research  and  Information Center (HWRIC), the Department of Energy and Natural  Resources,
Champaign, Illinois; the Civil Engineering Department at the University of Illinois; and the USEPA Office
of Research and Development,  Cincinnati, Ohio.

Project Goals and Objectives
       The purpose of this project was to evaluate potential technologies and operational modifications
that could reduce the amount of hazardous waste generated at a metal fabrication facility.  HWRIC's
pollution prevention team worked with R.B. White, Inc., a sheetmetal fabricator, to develop a reduction
strategy for hazardous waste generated from operating a 5000-gallon iron phosphating/degreasing
bath.  The goal of this project was to find an environmentally responsible means to extend the life of
the bath and thereby reduce hazardous waste generation. This project was carried out in four stages:
(1) initial assessment of the problem  and evaluation of alternatives,  in which ultrafiltration was
identified as the most promising technology to accomplish the project goal, (2) bench-scale screening
of ultrafiltration membrane candidates, (3) pilot-scale testing performed at HWRIC labs, and (4) full-
scale  implementation  and testing performed  on site at R.B.  White.    The  relative feasibility of
ultrafiltration as well as its capability to reduce waste generation were assessed on an engineering and
economic basis. Results of this project were used to justify installing a permanent ultrafiltration system
and operating practices that would improve product quality.

Industrial Participant
        R.B. White, Inc. operates a sheetmetal fabrication facility that manufactures painted  steel
shelving units.  Figure 1 depicts a floor plan of R.B.  White's  metal working, cleaning, and painting
operations. Cold-rolled steel arrives at the plant from the steel mill coated  with mill oils to protect the
bare metal from corroding or staining during storage  and fabrication operations.  The sheetmetal  is
fabricated using various stamping and shaping operations which apply coolants and lubricants to the
metal working surface.  Before being painted, the metal surfaces are cleaned to remove the mill oils
and metal working fluids and then preconditioned to bond well with the paint coating.  Fabricated parts
are  hung  on a  conveyor  system  that  immerses  the  steel  in   a   heated,  aqueous   iron

-------
Conveyor  j
 System  ;
          I
 parts
loading
     Parts
     Flow !
                                        X
                             Paint Spray ^
                             Booths    \
                                                           Infrared Oven
                                                            Curing
unloading
  parts
                                >_Air_Dryjng_
                                "funnel"
                                               I
                                                                     Finished Product
                                                                 Inspection and Packaging
                                 (D Spray
                                  Rinsing
                                      Iron Phosphating/
                                     Degreasing Bath
                                                   Stamping and Shaping Operations
                                    Ultrafiltration
                                        Unit
                      Path of Fabricated Sheetmetal
              Figure 1:  Floor Plan of Cleaning and Painting Operations

-------
phosphating/degreasing bath.  Surfactants in the bath clean away the mill oils, coolants, lubricants,
and shop soils while phosphating agents simultaneously deposit an amorphous coating that aids in
paint adhesion. Quality control in the bath is maintained by performing daily titrations which track the
level of chemicals in the bath. When the parts emerge from the bath, a fresh water spray rinses away
any oil residue and helps to prevent spotting.  Parts are visually inspected to make sure they are free
from oil and display a uniform phosphate coating.  The cleaned and phosphated parts are dried in a
forced-air drying tunnel, coated with enamel paint in a spray booth, and cured in an infrared oven. The
parts are then removed from the conveyor, and the adhesion quality of the paint coating is spot-
checked  using a cross-hatch test.  Finally, the finished product is boxed and shipped for distribution.
There are various point sources of waste generation in this process. Figure 2 illustrates a flow diagram
of the main operations and their associated waste  streams.   This project focused  on the hazardous
waste generated from  operating the 5000 gallon iron phosphating/degreasing bath and the resulting
rinse water discharge.
       R.B. White has used the current mode of iron phosphating/degreasing for the past eight years.
This process was recommended by DuBois Chemicals (R.B. White's chemical supplier) on the basis that
phosphating and degreasing could be accomplished in a single tank by using two of their products:
Dura-Gard Soke and Tart liquid acid.  Previously, the facility had operated separate degreasing and
phosphating tanks using trichloroethylene in the degreasing tank. In 1985, the company switched to
DuBois'  aqueous iron phosphating/degreasing  system  to improve worker  safety and reduce the
generation of organic solvent emissions and hazardous waste. Although switching to a single-stage
aqueous system eliminated the risks and liabilities associated with organic solvents, it introduced a new
waste disposal problem.

Problem Description
       At R.B. White,  simultaneous degreasing and phosphating in the same bath formed an oil-water
emulsion. With  extended use, the buildup of oil and dirt in the bath reduced cleaning and phosphating
efficiency, and  product quality  was compromised.  Without a clean surface, residual oil  and dirt
adversely affected the adhesion and uniformity of the phosphate coating.  Additionally, dragout of oil
from the bath into the rinse water raised  oil and grease levels in the discharge.  In the  past,  oil
skimmers were  used to control oil slicks on the surface and prolong the life of the bath, but the
skimmers were only partially effective.  When oil in  the bath began to sacrifice product quality and the
discharge levels edged closer to the maximum allowable limit, the bath had to be replaced. Depending
on production rates, the bath typically lasted 3 to 4 months. Replacing the bath required a full day of
lost production time to take the process off-line, make arrangements with a waste transporter to drain
and dispose of the entire contents, and recharge the tank with 5000 gallons of fresh water and raw

-------
        WASTE
    PRECURSORS
     PROCESSES
                          MAIN WASTES
                           GENERATED
   Oil-based rust
   inhibitor
   Metal working fluids
   Shop soils
Delivery of cold-rolled steel
  Stamping and Shaping
   Surfactants
+  Phosphating chemicals
4-  Water
                            Scrap metal
    Iron Phosphating/
       Degreasing
   Fresh water
    Enamel paint
    Cleaning solvents
                         h* Waste aqueous bath
                            with emulsified oil & dirt
                         •t* Iron phosphate
                            precipitate sludge
                         »:* Free oil
      Spray Rinsing
                              Forced-air Drying
Spray Enamel Painting
 K  '             a
                                Oven Curing
                         •:*  Rinse water
                             (discharged)
                             *  Spent cleaning
                                solvents
                             <•  Overspray to |3aint
                                filters (landfilled)
                                                          Solvent vapors
                                                          (incinerated)
   Figure 2: Flow Diagram of Main Operations and Associated Waste Streams

-------
materials.  The spent bath was classified as RCRA  hazardous  waste because it failed Toxicity
Characteristic Leaching Procedure (TCLP) tests for xylene.  Since land disposal of liquid wastes is
prohibited, the bath,  sludge, and skimmed oil were incinerated in a cement  kiln.  Disposal costs
including transportation and incineration ran about $1 per gallon which came to $5000 per bath, or
about §15,000 per year in addition to the costs associated with lost production time and replacement
of water and raw materials.

-------
                                          Section 2
                                  PROCESS BACKGROUND

       Iron phosphating/degreasing processes are widely used in the manufacture of metal products
to clean and precondition ferrous surfaces. Many metal fabricators and others that paint or coat steel
choose iron phosphating/degreasing processes because they effectively clean metal parts, provide an
excellent surface for paint adhesion, and  protect against under-paint corrosion (Wittke, 1987).  Iron
phosphate coatings have become increasingly popular in prepaint applications as existing and projected
EPA regulations require more stringent control and more expensive waste treatment methods for other
coatings such as chromated seals, zinc phosphate, and chromate conversion coatings {Phillips, 1990).
The goals and mechanisms  associated with  R.B. White's degreasing and phosphating process are
discussed below.
       The goal of degreasing is to remove mill oils, metal working fluids, and any other shop  soils
from the steel surface and prepare it for finishing.  Degreasing agents can accomplish this in several
ways: detergency, solvency, chemical reaction, or mechanical action (USEPA, 1989).  The aqueous
cleaning solution in use at R.B. White employs detergency in a heated {140°F), air-agitated bath.
        In  R.B. White's bath, degreasing is accomplished using nonionic  surfactants that act  at
interfaces between the metal, oil, dirt, and water. The surfactant molecules have hydrophobic "tails"
that penetrate the oil and dirt while their hydrophilic "heads" surround  the particles and lift them away
from the  metal  surface and into the water phase.  Nonionic surfactant molecules owe their "split
 personality"  (Wray,  1992) to long chain, hydrocarbon tails that more readily solubilize in oil than in
 water and to oxygen-containing heads that rather prefer to  hydrogen-bond with water molecules.
 Surfactant molecules  arrange themselves  in  chemically  stable micelle  formations, with  their
 hydrophobic tails  oriented inward and their hydrophilic heads pointing out.  The micelles disperse in
 the water phase and form a stable emulsion that cleans the parts and prevents the oil and dirt from
 redepositing on the metal surface (Morrison,  Boyd, 1973).  Micelles form as  aggregates of 50 to 300
 or more  surfactant  molecules depending on  the nature, temperature,  and  concentration of the
 surfactant.  Micelle formation  will only occur at  or  above  a  very  narrow range  of surfactant
 concentration  known  as the  critical  micelle  concentration  (cmc).   Below the critical  micelle
 concentration, there is not enough surface activity to provide  efficient cleaning (Cutler, Davis, 1972).
 Studies by Ginn and Harris (1961) have shown that maximum  soil removal for nonionic surfactants
 occurs at concentrations well above the critical micelle concentration. In  R.B. White's degreasing
 application, maintaining sufficient levels of surfactant in the iron phosphating/degreasing bath is an
 important process control parameter to ensure good cleaning.

                                               6

-------
       Phosphating is a type of prepaint coating process used to simultaneously provide corrosion
                                     -%•  i'"^          '-;ft.  if*'
resistance and enhance paint adhesion to a  metal surface.  Phosphoric acid chemically bonds to the
metal surface to produce an amorphous conversion coating.   The phosphate conversion coating is
nonconductive so it protects the metal surface from electrochemical oxidation that leads to rust and
corrosion. The matrix of the phosphate coating forms capillaries which increase the surface area and
provide a mechanical interlocking structure on which the paint can adhere (Phillips, 1990).  Since the
phosphate  layers contain surface breaks, immediate  painting  is imperative or corrosion may begin
within twenty-four  hours following the phosphating operation (Paul, 1986).
       The phosphate coating process chemically reacts in two steps.  Phosphoric acid is added to
lower the bath pH  to between 3.5  and  5 to dissolve a thin layer of metal from the part surface.
Solubilizing the metal consumes acid, and the local pH at the metal surface increases.  At this higher
pH, phosphate salts bond with the solubilized metal ions to form an insoluble phosphate precipitate.
The  precipitate deposits  on the metal surface in  an amorphous configuration.   The two critical
parameters  in  this  process are the solution pH and  the  concentration of  phosphating agents.
Maintaining a solution pH between 3.5 and 5 is critical to ensure optimum reaction between the metal
surface and the phosphoric acid.  Keeping the concentration of phosphating agents sufficiently high
is important to provide adequate coating  and coverage, but excessive concentrations may produce a
coating that is undesirably loose and dusty (Phillips, 1990).
        Parameters  describing the quality  of an iron phosphate coating include:  (1) coating weight, (2)
rust creepage, and  (3) paint adhesion. Coating  weight is a measure of the mass of iron phosphate
(Fe3(PO4)2) deposited per unit area. Rust creepage is an indication of coating density and uniformity.
Paint adhesion is a measure of the  strength of  the bond between the phosphate  and paint layers.
Imperfections in the coating can usually be traced to fluctuations in pH, changes in temperature, or an
imbalance of chemicals or contaminates in the bath (Phillips, 1990).
        Several system  configurations are  available  for iron phosphating/degreasing  processes.
Systems may utilize vapor, spray, or immersion  applications and range anywhere from one to seven
steps. Process designs may include  a pickling step to remove rust and scale. Systems also have the
option of using distilled water  or applying a seal coat in the final rinse. Selection is determined by
product quality requirements, soil removal capability, and production  load.  Traditionally, chemical
suppliers have recommended separate degreasing and iron phosphating steps when soils were difficult
to remove or when cleanliness was a particularly important quality parameter. Manufacturers would
rely on solvents or aqueous alkaline cleaners to degrease the parts, fresh water to rinse, iron phosphate
solution to coat, and more water to rinse.  In many applications, adequate product quality can be also
be  achieved by using a system configuration that combines  the  degreasing and  iron phosphating
process into one step.  This  configuration  relys on surfactants to function concurrently  with iron

-------
phosphating agents in an aqueous solution. Companies like R.B. White switched to the one-step iron
phosphating/degreasing bath because it offered an alternative to solvent use and disposal, eliminated
separate cleaning and rinsing stages, and reduced water and space usage.
       The 5000-gallon bath at R.B. White was charged with Dura-Gard Soke and Tart liquid acid
which contained nonionic surfactants, phosphate salts, phosphoric acid, and accelerators to promote
phosphate precipitation.  The concentration of Dura-Gard Soke was checked daily with  a simple
titration kit and conversion table to ensure that concentrations were maintained between 1.5 and 2.0
ounces of Dura-Gard per gallon. After the concentration of Dura-Gard was adjusted, pH was checked
with litmus paper. If the pH exceeded 3.5, it was adjusted to 3.5 by addition of Tart liquid acid. The
bath temperature was maintained at 140°F.  During operation, a coarse bubble  aeration  system was
used to prevent bath stratification and provide complete mixing of all but the sludge settled on the tank
bottom.  Oil skimmers were used to control oil slicks on the surface.  Daily operations varied in length
from  8 to 16 hours per day, 5 to 6 days per week.
                                               8

-------
                                         Section 3
            WASTE REDUCTION IN THE METAL FABRICATED PRODUCTS INDUSTRY

Waste Generation
       The EPA's campaign for waste reduction is bringing change to industries through the 1984
Hazardous and Solid Waste Amendments (HSWA) to the Resource Conservation and Recovery Act
(RCRAj, the Toxics Release Inventory  (TRI), the 1990 Pollution Prevention Act (PPA), and the more
recent 33/50 Program.  The HSWA require industries to set up waste minimization programs and to
produce certified manifests demonstrating their waste reduction efforts. The TRI is a computerized
data base that tracks the routine and accidental release of approximately 300 toxic chemicals reported
by U.S. manufacturers. The 1990 PPA brought about stricter TRI industrial reporting requirements that
include providing information on pollution prevention efforts. The 33/50 Program is EPA's voluntary
pollution prevention initiative to reduce the Nation's releases of 17 TRI chemicals 33% by the end of
1992 and 50%  by the end of 1995 (Hindin, Burch, Fort, 1992).  Backed by federal legislation and
economic incentives, EPA's pollution prevention campaign has targeted several operations associated
with the metal fabricated products industry-  Finding environmentally responsible solutions to the
industry's waste disposal problems has focused on source reduction (including process modifications
and raw materials  substitution) and recycling.
       The metal  fabricated products industry is an integral part of aerospace,  electronic, defense,
automotive, furniture, domestic appliance, and many other industries. The raw materials used in metal
fabrication range from common copper and  steel to high grade alloys and precious metals.  Typical
materials used and types of wastes generated from metal shaping, surface preparation, and surface
finishing operations are  presented in Table 1.
       Wastes generated from metal shaping operations  include scrap metal and metalworking fluids.
In general, metal working fluids are petroleum-based fluids, oil-water emulsions, or synthetic emulsions.
With extended use, these metalworking fluids become spoiled and are often disposed as  hazardous
waste.  The fluids may  be contaminated with cadmium, chromium, or lead depending on the metal
being tooled.   In addition, the fluids  may contain  chemical additives such  as  chlorine, sulfur and
phosphorous compounds,  phenols, cresols, and alkalies (Hindin, 1992).
       Surface preparation operations generate wastes contaminated with solvents, heavy metals, and
oils.  Concentrated solvent-bearing wastes may result  from degreasing operations which produce
solvent-bearing wastewaters, air emissions,  and solid-phase wastes.  Chemical treatment operations
such as alkaline, acid, mechanical, and abrasive cleaning can generate waste streams contaminated
with oils or heavy metals (Hindin,  1992).

-------
       Surface finishing operations result in plating-related wastes and painting wastes. Metal plating

and related metal deposition processes such as conversion coating account for the largest volume of

heavy metal- (cadmium, chromium, lead mercury, nickel) and cyanide-bearing wastes. Painting and

cleanup operations result in the direct release of solvents such as MEK, M1BK, toluene, xylene, and

many TRI regulated chlorinated solvents (Hindin, 1992).


Table 1:  Typical Materials Used and Types of Wastes Generated in the Metal Fabricated Products
Industry (Hindin, Burch, Fort, 1992)
  Operation
Typical Materials Used
Types of Wastes Generated
  METAL WORKING
 Cutting fluids
 Lubricants
 Inorganic/organic solvent
 electrolytes
 Heavy metals
. Heavy metal wastes
. Solvent wastes
. Waste oils
. Scrap metal, filings
  SURFACE
  PREPARATION
  Cleaning
 Acid/alkaline cleaners
 Detergents
 Organic degreasing and
 cleaning solvents
 Acid/alkaline wastes
 Ignitable wastes
 Solvent wastes
 Still bottoms
 Waste oils
  Chemical Treating
  Acid/alkaline cleaners
  Etching/Pickling acids
 Acid/alkaline wastes
 Heavy metal wastes
 Waste oils
  SURFACE FINISHING
  Electroplating
  Acid/alkaline solutions
  Heavy metal-bearing solutions
  Cyanide-bearing solutions
  Acid/alkaline wastes
  Cyanide wastes
  Heavy metal wastes
  Plating wastes
  Reactive wastes
  Wastewaters
  Conversion Coating
  Phosphate salts
  Chromate salts
  Acid solutions
  Chromium-bearing
  wastewaters
  Acid wastes
  Painting
  Paints
  Solvents
  Paint carrier fluids
  Heavy metal paint wastes
  Ignitable paint wastes
  Solvent wastes
  Still bottoms
  Paint overspray waste
  VOC emissions
 Aqueous Cleaners

        The  EPA recommends various strategies for pollution prevention in the metal fabricated

 products industries.  Until now,  waste reduction in surface preparation operations has focused on
                                              10

-------
conserving or finding alternatives to organic solvent cleaners. For years, the metal finishing industry
has relied on  organic solvents for cleaning metal parts. 
-------
parts.  In one case study, a major U.S. aerospace manufacturer successfully made the switch from a
1,1,1-trichloroethane vapor degreaser to an aqueous cleaning bath with only slight modifications and
little capital investment. In another case, a large plating firm replaced a two-stage solvent degreaser
and caustic bath with a single-stage, heated (140°F), aqueous cleaning bath (Quitmeyer,  1991). In
1985, R.B. White also made the switch from trichioroethane to a single-stage, heated (140°F), aqueous
degreasing/phosphating system and eliminated separate degreasing and rinsing operations.
       In all these applications, the tank life of the aqueous cleaners is limited by the buildup of dirt
and oil in the bath. Cleaning effectiveness begins to deteriorate as dirt and oil accumulate in the water,
and the performance of other chemicals in solution is inhibited. Although the aqueous degreasers do
not carry the risks and liabilities associated  with the  disposal of waste organic solvent cleaners,
periodic replacement of the bath creates a different waste disposal problem.  Aqueous degreasers form
stable oil-water emulsions that work well in cleaning but make them difficult to treat for disposal.  The
emulsions contain levels of dirt, metal particles, and oil that are not acceptable to pour down the drain.
When pretreatment  limits for wastewater discharge are exceeded, high oil content  and chemical
oxygen demand (COD) keep the waste process  solution from  being treated at a publicly owned,
treatment works (POTW) by conventional biological treatment methods {Bailey, 1977).
        Current  disposal options for spent aqueous cleaning solutions include tankering to an offsite
treatment facility, incineration on or offsite, or discharge to a POTW. Tankering the waste process
solution to another site is a convenient but expensive  option since more than 90% of the waste is
actually water.  Tankering is also not a pollution prevention solution since it only transfers the problem
from one location to another.  Incineration is another disposal option that can be used to burn the
waste oil despite its high water content.  Incineration is commonly used, but the process requires
supplemental fuel and leaves no room for reuse or recovery of the spent bath (Bailey, 1977).
        Discharge to a sanitary sewer is another option for disposal only as long as the wastewater
meets with  effluent standards.   An  industrial facility may  operate  an  end-of-pipe, wastewater
pretreatment  plant  before  discharging  spent cleaning solution to the  POTW.   A  conventional
wastewater pretreatment plant might employ the following unit processes to remove oil from water
prior to discharge.  Skimming can remove free oil from the surface, but sulfuric acid or lime may be
required to adjust the  pH and help release free oil. Coalescing can separate  water and dispersed oil
droplets that do not float to the surface, but this technique is only effective when surfactants are! not
present.  Additional chemicals are required to overcome surface active agents and release the: oil.
 Chemical emulsion breaking can be used when coalescing is not effective. Chemical emulsion breaking
 begins by adding aluminum sulfate (alum) as a coagulant to destabilize the emulsified oil droplets. Lime
 or caustic soda is used to raise the pH and form an  insoluble aluminum  hydroxide floe.  Cationic
 polymers may also be used with alum to eliminate the need for pH  adjustment.  The destabilized oil
                                              12

-------
droplet:; then adsorb onto the alum floe surface. Dissolved air flotation (DAF) employs tiny bubbles
of air to carry the alum floe to the surface and over a clarifier weir (Springborn Laboratories, 1982).
       The primary goal of this type of conventional, end-of-pipe pretreatment is compliance. Treated
water is discharged to the sewer while sludge produced during the process creates a new waste
disposal  problem.   With increasing  environmental  restrictions,  landfilling  the chemical sludge is
becoming less and less attractive as a disposal option.  Alternatively, the sludge can be incinerated or
treated on site to recover the oil.  Thermal oil recovery relies on heating the waste sludge to nearly the
boiling point to separate the oil, water, and sludge phases. The process may require a heat exchanger,
centrifuge, evaporator, acid, and additional chemicals (Springborn Laboratories, 1982). Since waste
emulsified oil  solutions typically contain less that 10% oil, it usually does not pay to install such an
extensive oil recovery system (Cheng, 1983).
       The rising costs associated with these disposal and pollution control techniques are the two
main incentives to extend  the life of the aqueous cleaner baths (Wahl, 1979).  Rather than wasting
valuable raw  materials, the aqueous cleaners have the potential  to be recycled again  and again.
Depending on the physical characteristics of the bath solution, the life of the bath can be extended by
skimming contaminants off the top, settling heavier fractions to the bottom, or filtering out suspended
species (Quitmeyer, 1991).  R.B. White found skimming only marginally effective,  and  no amount of
settling time would break the stable oil-in-water emulsion, but filtration did have potential for extending
the life of their aqueous cleaning bath.

Ultra-filtration
        Conventional filtration techniques rely on depth or screen filters to remove oil and dirt from a
process solution, but conventional filter media clog easily. They require frequent  backflushing or
disposal which result in additional wastes.   Membrane  filtration  is a more advanced technique that
takes advantage of thin-film membranes and turbulent flow patterns to deliver a more consistent flow
rate and a higher quality filtrate than conventional filtration.  One way that membrane filtration differs
from conventional filtration is that the direction of feed flow is tangential to the filter media rather than
perpendicular to it. Turbulent cross-flow across the membrane means oil and dirt are less likely to
 build-up  on the surface of the filter media and clog  the  flow (Bailey,  1977).  The  membranes
themselves are also revolutionary. Instead  of thick beds of filter media, membranes are cast as thin
 films with designated pore sizes ranging from atomic to macroscopic levels. Ultrafiltration  is one class
 of membrane filtration that uses membranes with pore diameters ranging from 10~9 to 10"8 meters.
 This is smaller than the size of an emulsified oil droplet but much larger than the water molecules and
 other dissolved species in an aqueous cleaning solution  (Cheryan, 1986).
        The ultrafiltration  process works by producing two  separate streams:  concentrate  and
                                               13

-------
permeate. The permeate stream contains only the components in the feed solution small enough to
pass through the membrane pores.  For an aqueous cleaning bath like R.B. White's, the permeate
would contain water, phosphating agents, and solubilized raw materials that did not react with the
membrane surface. The concentrate stream contains everything else that is rejected by the membrane.
At R.B. White, this would include emulsified oils, dirt, and water associated with the emulsion.
       The recent development of more durable membranes, such as polyvinylidene di fluoride (PVDF),
has expanded the application of ultrafiltration beyond its origins in the food industry to  successfully
handle industrial process solutions with extreme pHs, high temperatures, and high oil concentrations.
Because  of  its  unique  capabilities to concentrate oily  wastewater  and  produce  a clear filtrate,
ultrafiltration has emerged as a promising technology for extending  the life of aqueous  cleaners.
Ultrafiltration of oil-water emulsions is a more straight-forward method for removing and concentrating
oil are than any of the chemical or thermal techniques mentioned earlier. Ultrafiltration does not require
a stockpile of chemicals such as sulfuric acid, alum, lime, caustic soda, polymers, and other proprietary
chemicals.  Instead, ultrafiltration produces a water phase that requires no further treatment and a
concentrated phase only a fraction of the original volume that can sustain combustion or be disposed
of efficiently.  Ultrafiltration requires no heat input, low energy, and little  operator attention (Wahl,
1979).
       Each ultrafiltration application calls for a specific  membrane type.   Membranes used in
ultrafiltration are characterized by material of construction, pore size statistics, rejection, and flux.  The
material of construction  determines the  acceptable range of  chemical  and physical parameters
(resistance to pH, temperature, pressure, solvents) that the membrane can withstand. Pore  size
statistics describe the membrane's structure in terms of nominal  pore size,  pore size distribution, and
percent porosity- These parameters determine the minimum molecular weight cut-off (MWCO) which
is the minimum weight of a typical molecule that would be rejected by the membrane. Rejection
describes the degree to which ultrafiltration prevents components from passing through the membrane.
The rejection may change with use or aging of the membrane,  but the % rejection may always be
determined  by the following equation:
               % Rejection = (CF - CP)/CF X 100%                                      (E!q. 1)
               where CF =  concentration in feed solution
                      CP =  concentration in permeate  stream
 Flux is defined as the volumetric flow rate of permeate  per cross-sectional area per time.  Flux is  a
 function of pressure, feed concentration, temperature, flow rate and turbulence in the feed channel,
 and the tendency to foul (Cheryan, 1986). Rux determines the membrane surface area and pump
 capacity requirements for a given wastewater flow rate or volume. Therefore, capital and operational

                                               14

-------
costs are directly related to flux characteristics.
       Membranes are cast in a variety 6f module configurations, each having particular advantages
and limitations.  Spiral wound,  hollow fiber, and tubular membranes are the most popular cross-flow
configurations. These configurations are designed to supply the greatest amount of surface area per
unit volume and flow channels  that will accommodate the nature of the feed stream.  Illustrations  of
each  are provided in  Figure 3.  The wide-channel tubular configuration is  usually preferred for high
strength oily wastewaters. Although tubular membranes do not have the high surface area to volume
ratios that the other configurations do, the wide-channel tubular membranes provide good resistance
to cake formation and clogging by oil and particulates (Cheryan, 1986).
        One of the greatest limitations of ultrafiltration membranes is their tendency to clog or foul.
Fouling is detected as the decrease in permeate flux over time and is a result of changes in the
membrane structure or interactions between the components of the feed stream and  the membrane
surface.  Fouling  is mainly  due to  the accumulation of particles and adsorption of foulants on the
membrane surface and/or within the pores of the membrane itself.  Almost all feed solutions will foul
a membrane to some extent, but optimizing the chemical compatibility of the feed stream and the
membrane material will  help to reduce the  effects of fouling.  When a membrane shows signs  of
fouling, the flux can largely be restored by cleaning the membrane to remove accumulated foulants like
oil, suspended solids, and metal precipitates.  Mechanical cleaning employs sponge balls to remove
chemically precipitated species. Detergents work to breakup deposits on the membrane surface and
disperse them into solution. Solubilizing uses acids and chelating agents to dissolve stubborn foulants
like metal hydroxides or other chemical deposits {Pinto, 1978).  Even with  regular cleaning, a portion
of the flux may be unrecoverable because of irreversible fouling.  The extent of irreversible fouling
determines the useful life of the membrane, which is also part of the operational cost associated with
the process (Cheryan, 1986).
        The rate of fouling  depends not only  on the  type of membrane and the nature  of the feed
stream, but on the operating conditions as well. Fouling can be minimized by controlling pretreatment,
temperature,  pressure, velocity, configuration, and concentration.  Pretreatment of the feed stream
 protects  the membrane  surface by straining out debris like suspended solids, grit, metal filings, and
 cigarette butts.  Temperature controls the viscosity of the feed stream and the adsorption of foulants
 to the membrane surface. Increasing temperature within the recommended  range for a membrane may
 increase  flux and decrease foulant adsorption. Transmembrane pressures are usually specified by the
 membrane manufacturer to deliver optimal flux without compressing  or changing  the membrane
 structure.  Increasing the pressure within a certain  limit will yield an increase in flux.  Velocity in the
 flow channel determines the degree of turbulence and, therefore, the likelihood of cake formation at
 the membrane surface.  High velocities at the membrane surface tend to shear off deposited material
                                              15

-------
                                 t'UKMEATE
                      -EPOXY REINFORCED                    /-BOOT SEAL
    MEMBRANE--          FIBERGLASS SUPPORT TUBE

  Figure 3a:  Tubular Membrane Module (Cheryan, 1986)      !
                                                          CONCENTBATEOUT
   FEED IN
                 PERMEATE
                   SPACER
                              MEMBRANE     FEED SPACER
  Figure 3b:  Spiral Wound Membrane Module (Cheryan, 1986)
                                                     OPEN END
                                                     Of FIBERS
                                                                EPOXY
                 TUBE SHEET
      SNAP RING

XT RING SEAL
    \


 FEED
                               aow SCREEN
                       POROUS
                      BACK-UP DISC
                           SNAP RING
                                                                          PERMEATE
     END PLATE
                        FIBER
                                   SHELL
            XT RING SEAL
  POROUS FEED          END PLATE
DISTRIBUTOR TUBE
          CONCENTRATE
   Figure 3c:  Hollow  Fiber Membrane Module (Donnelly, Goldsmith,
                                                    McNulty, 1976)
                                          16

-------
and prevent the formation of a fouling layer.  Membrane configurations establish flow patterns to
minimize fouling from different types of feed streams.  Choosing the best membrane configuration will
depend on the nature of the feed  stream, the type of application, and the  economies of scale.
Concentration of the feed stream determines the rate of permeate  flux and the potential for  cake
formation at the membrane surface. While it is possible to achieve 40 - 75% oil in the concentrate,
it may be more economical to settle for 40 - 50% since the rate of permeation decreases rapidly as
feed stream concentration rises. Higher feed stream  concentrations  also have,a greater potential to
foul the membrane, which leads to higher membrane cleaning and  replacement costs (Bailey, 1977).
       In industrial applications where ultrafiltration could be used to filter aqueous cleaning baths,
fouling will typically be due to oils, suspended solids, free surfactants, and metal precipitates (Pinto,
1978),, Research by Bhattacharyya, et. al. (1979) predicted fouling phenomena for oil-surfactant-water
conditions similar to those at R.B. White.   Bhattacharyya experimented with  ultrafiltration of oily-
surfactant waters using tubular, non-cellulosic membranes.  The  study was used to determine the
feasibility of using ultrafiltration aboard Navy ships to treat "bilge waters" to meet marine discharge
standards. The bilge oil Bhattacharyya used to mix the samples consisted mainly of fuel oil, lubricating
oils, and hydraulic oils.  The experimental formulations were prepared at a neutral pH using distilled
water or river water and nonionic surfactants.  Bhattacharyya worked at temperatures of 25 and 40°C
(77 and  104°F) and tested different combinations of feed streams: oil-water, surfactant-water,  and
oil-surfactant-water.
        From the test results, Bhattacharyya attributed fouling and flux decline to a cpmbination of gel
formation and  adsorption of foulants on the membrane surface.  In the oil-river water samples,
Bhattacharyya found that the interaction between oil and suspended solids resulted in the formation
of a gel layer that restricted permeate flux.  Gel formation occurs when the interactions between the
oil droplets are strong enough to produce aggregates that eventually develop into a gel layer on the
membrane surface (Lee, 1984).  With the dirt and other particulates  in  R.B. White's bath, the formation
of a gel layer was a possible source of fouling. In the surfactant-water samples, Bhattacharyya found
that free nonionic surfactants contributed to fouling and inhibited flux by physical adsorption  and
micelle formation within the membrane  pores.  In  R.B. White's application, this type of surfactant
stripping by the membrane was a concern for preserving the concentration of raw materials in the bath.
In the oil-surfactant-water systems,.Bhattacharyya  found that there was less flux decline than in the
other two systems. With the surfactant and oil tied up in an emulsion, there was decreased adsorption
of free surfactant and oil on the membrane surface.  However, since less free surfactant was available,
surfactant rejection was enhanced in the presence of oil. This scenario most closely represented the
actual conditions at the R.B. White facility.  In general, the degree of adsorption depends on the type
and concentration of surfactant, other materials present in solution, and the type of adsorbent  (oil,
                                             17

-------
membrane, etc.).  Adsorption of surfactant  tends  to decrease  as temperature  increases and  as
concentrations  rise above the critical micelle concentration (Cutler, Davis, 1972).   By raising the
temperature to 40°C (104°F), Bhattacharyya found that flux increased due to lower viscosities and less
adsorption of oil and surfactants on the membrane without sacrificing oil rejection (Bhattacharyya,
Jumawan, Grieves, 1979).   For  the  warmer operating temperatures (140°F) at R.B. White, this
observation was encouraging.
                                              18

-------
                                         Section 4
                           BENCH-SCALE MEMBRANE SCREENING

Methods and Materials
       The laboratory bench-scale activities served as an initial screening of candidate membranes to
asses their respective flow rates  and chemical compatibility.  The key  parameters for membrane
evaluation focused on flux and fouling characteristics and permeate quality.
       Ultrafiltration equipment used in this study was provided by a variety of well-known equipment
and membrane manufacturers. The laboratory bench-scale study utilized an Amicon dead-end, stirred-
cell ultrafiltration assembly (Model 8200, 180 mL capacity) with a nitrogen pressure  source.  The
bench-scale apparatus was used to screen a variety of 62 mm diameter disk membranes for flux rates
and chemical compatibility. A summary of the disk membranes tested is presented in Table 2, and a
diagram of the bench-scale apparatus appears in Figure 4.  Permeate flux rates were determined by
weighing the permeate  as it filtered through the stirred cell over time.  Permeate was collected in a
beaker located on a Sartorius analytical balance (Model LC 12000s). The balance was  connected to
a computer via a RS-232 cable,  and mass measurements were sent in prescribed time  intervals to a
Lotus 1-2-3 spreadsheet.
Table ?.:  Membrane Candidates and Material Specifications
MODEL
NO.
HG05
SNO4
PM10
PM30
M2
M3
VENDOR
Osmonics
Osmonics
Amicon
Amicon
Enviro-Process
Enviro-Process
MATERIAL OF
CONSTRUCTION
Polysulfone
Cellulose Acetate
Polysulfone
Polysulfone
Teflon
Teflon
MWCO
2,000
20,000
10,000
30,000
20,000
500
MAXIMUM
PRESSURE
(psi)
400
150
70
70
80
200
MAXIMUM
TEMPERATURE (°C)
100
60
100
100
* No limit specified
* No limit specified
Bench-Scale Activities
       Samples for the bench-scale study were collected from the iron phosphating/degreasing bath
while it was relatively dirty. Samples were drawn from twelve locations in the tank along the surface,
middle, and bottom using a Scienceware polyethylene dipper.  The dead-end, stirred ultrafiltration cell
                                             19

-------
   Pressure from
   Nitrogen Source
Disk Membrane
Magnetic Stirrer
                                         Permeate
                                                                Beaker
                                                 Electronic Balance
                                                   Computer B
                                                               Data via
                                                               RS-232 Cable
    Figure 4: Bench-Scale Ultrafiltration Apparatus
                (Laine, Hagstrom, Clark, 1989)
                                  20

-------
was used to test six different membranes for clean water flux, process solution flux, and flux recovery
at consistent transmembrane pressures in triplicate. First a clean water flux test was run with  180 mL
distilled/deionized water to establish an initial flux. The process solution flux test directly followed to
characterize a representative flow rate of process solution through the membrane. Flux recovery was
determined by rinsing the membrane and filtering a second 180 mL aliquot of distilled/deionized water.
The deterioration of process solution flux  with time and  the net difference in initial flux and flux
recovery were used as a measure of fouling that occurred during filtration.
       Sample quality for the bench-scale testing was defined by the partition-gravimetric method for
oil and grease (USEPA Method 413.1). At this stage,  oil and grease tests were performed without
sufficient analysis to actually differentiate between oil/grease and non-ionic surfactants because the
contract lab used was unaware of the importance of specifying  these components separately (see
Quality Assurance section).  Oil and grease analyses were  performed on permeate, concentrate, and
unfiltered process solution samples. Results are still considered valid to the extent that they were used
to determine the relative efficiency of each membrane to remove emulsified oil  contamination in the
process solution and last without fouling.

Results and Discussion
       The results of the bench-scale flux tests  are summarized in Figures 5a and 5b.   Figure 5a
compares the relative wastewater flux  rates of each membrane candidate. Model M2 (Teflon 30,000
MWCO) produced the highest flux rates with Model SN04 (Cellulose Acetate 20,000 MWCO) running
a close second.  Figure 5b shows that both candidates  displayed competitive flux  recovery, but
         iai  values set the M2  Teflon model out front.  J represents the permeate flux, and the value
      aae/Jiniti.1 "s an indicator of the loss in permeate flux from clean water to wastewater. The relative
loss was much less for the M2 Teflon model than the SN04 Cellulose Acetate which indicates the M2
Teflon membrane underwent the least  fouling in the bench-scale test.
        Oil  and grease  levels  in the  permeate as measured by the  USEPA  Method 413.1  were
consistently low, typically less than 40  mg/L for all membrane candidates. As discussed in the Quality
Assurance section of this report, a portion of the oil and grease measurements was attributed to the
presence of free surfactant in  the permeate.
        The overall best performance was  given by the M2 Teflon 30,000 MWCO membrane from
Enviro-Process Systems.  Even though  the SN04 Cellulose Acetate membrane appeared to be a close
competitor, manufacturer's ratings indicated that the cellulose acetate would not have withstood the
extreme conditions at the R.B. White facility with long-term use. Attractive features of the M2 Teflon
model for this application were its resistance to extreme pH, its ability to withstand high temperatures,
and its durability in the presence of industrial wastewater components.
                                             21

-------
j:
E
"c
"2
o
E

1
 n
 E
 2
 x
 J3
 U.

 n
 to
 CO
 £
               M2

 MATERIAL: Teflon
                       SN04
                                 PM30
 MWCO:
  PM10

Polysulfone

10,000
                     Cellulose   Polysulfone   Teflon
                     Acetate
            30,000    20,000    2,000       500
                          Membrane Model
Figure  5a:   Comparison  of Wastewater  Flux  Rates
                During  Bench-Scale Testing
  X
  3
  •o
  O
  E
  o
         0.4-
         0.2-
                                                                 Javerage/Jinitial
                                                                 Jrecovery/Jinitial
                                                                 where J = flux
         0.0
                M2       SN04     PM30      M3      PM10
  MATERIAL:  Teflon     Cellulose   Polysulfone   Teflon   Polysulfone
                       Acetate
  MWCO:     30,000    20,000   30,000     500    10,000
                             Membrane  Model
  Figure 5b:   Comparison  of Normalized Flux Rates
                  During  Bench-Scale Testing
                                    22

-------
                                          Section 5
                                   PILOT-SCALE TESTING

Methods and Materials
       Although dead-end filtration is useful for bench-scale screening, pilot- and full-scale membrane
cartridges typically use hollow fiber, spiral wound, or tubular configurations (see Figure 3, Sect. 3).
Due to the high concentration of suspended  solids and other impurities associated with the process
solution, a tubular configuration was chosen for the pilot- and full-scale tests.  Although tubular
membranes do not have the high surface area to volume ratios that other configurations do, the wide-
channel, cross-flow nature of the tubular membranes provides good resistance to cake formation and
clogging by particulates.  Additionally, tubular membranes offer the advantage of being able to use
rubber sponge-balls in cleaning.  A pilot-scale, tubular module made of the Enviro-Process M2 Teflon
was not available, so other membrane suppliers were contacted to find a suitable replacement.  Koch
Membrane Systems' 5'HFP-276-PNO PVDF 100,000 MWCO membrane was recommended in the one-
inch tubular configuration  with  1.1 square feet  of membrane area.  Like Teflon, PVDF is capable of
withstanding high temperatures and extreme pHs. The PVDF 100,000 MWCO was greater than the
30,000 MWCO tested for the Teflon membrane,  but the PVDF surface was negatively charged to repel
emulsified oil droplets.  The larger nominal pore size would also permit higher permeate flux rates and
permeation of unused surfactants and phosphating agents.  The pilot-scale ultrafiltration unit utilized
the Osmonics PES/OSMO-O5T-SSXX model capable of operating at pressures between 40 and 60 psig
within a pH range of .5 to 13.
       Three types of large-scale operating schemes can be used in pilot- and full-scale applications:
continuous, batch, and modified-batch.  Scheme selection depends on the size of the operation, the
availability of space,  and the nature of the process solution.
       The continuous scheme implies a dead-end filtration configuration or a system with internal
recycle (see Figure 6).  Permeate  is returned  directly  to  the  bath as contaminants continuously
accumulate at the membrane surface and within  the system plumbing. This scheme does not bear the
additional capital cost of a separate holding tank for the permeate, but continuous or dead-end filtration
is likely to significantly shorten the useful life  of the membrane due to the rapid accumulation of
foulants.
        The batch scheme filters process solution one tank at a time.  Rgure 7 illustrates a  schematic
of the batch set-up.  When  process solution is  pumped from the feed tank through the UF system,
some of the flow permeates through the membrane into a separate tank while the majority of the flow
is recirculated back to the feed tank for further concentration.  As the flow is recirculated, more and
more solution is permeated through the membrane until only a fraction of the original solution volume
                                             23

-------
                  Feed Tank
                 (5000 gal bath)
                      I Pump

Figure 6: Continuous Scheme Ultrafiltration
                       Feed Tank
                     (5000 gal bath)
                                       Concentrate
               Pump •

 Figure 7:  Batch Scheme Ultrafiltration

                 Pump
              •£ nj "a?
              oj m c
              J= 0 0
              D. £ o
                t
                              Process Tank

                                 (55 gal)
Pump
                   ^Permeate to separate
                     holding tank
                                                Concentrate
                                                                   Membrane
                              Permeate to bath
 Figure 8: Modified-Batch Scheme Ultrafiltration
                                    24

-------
remains in the feed tank.  This type of set-up was tested in HWRIC's pilot lab facility using a 55-gallon
feed tank.  For a batch scheme to work on a full-scale operation, the entire  bath would have to be
filtered into another 5000 gallon  holding tank and then returned to the operating tank all at once.
Periodic batch filtration uses membranes more efficiently and helps to prolong the useful life of the
membranes. However, this option involves down time as well as the additional capital costs associated
with a separate holding tank and greater membrane capacity.
       The modified-batch  scheme is  a  cross between the batch and continuous schemes.  This
scheme utilizes an intermediate process tank which is only a fraction  of the size of the primary feed
tank. A schematic of the modified-batch scheme is depicted in Figure 8. The oily bath is first fed into
the process tank and then sent through  the ultrafiltration system. The clean permeate is continuously
returned to the bath while the concentrate  is recirculated back  to the  process tank  for further
concentration.  As the permeate is filtered into the bath and the process tank gets low, solution is
pumped from the bath into the process tank to keep it full.  When levels in the process tank reach
sufficient concentration, a batch-down  procedure can be used to further concentrate the contents of
the process tank. During a batch-down, flow to the process tank is stopped, and the contents of the
process tank are allowed to concentrate down as described for the batch scheme. The modified-batch
scheme was chosen in the full-scale testing for its cost efficiency, lack of disruption to the production
process, and limited space requirements.

Pilot-Scale Activities
       A batch  scheme ultrafiltration system was set up at HWRIC's pilot lab facility to filter bulk
samples  of process solution.  The pilot-scale system was  intended  to model the flux and fouling
characteristics that might be expected in a full-scale operation.
       A preliminary pilot test was conducted on clean process solution to determine the selective
permeation of raw materials through the  PVDF membrane.  Since the two essential components of
Dura-Gard Soke are phosphating agents and surfactants, it was important to monitor the fate of these
chemicals under the influence of ultrafiltration. Twenty gallons of clean process solution were prepared
with distilled water at 2.5 ounces Dura-Gard/gallon and filtered through the UF pilot unit.
       A second pilot-scale test was used to evaluate the effectiveness of the UF equipment regarding
removal of emulsified oil and raw materials from the waste process solution. A 50-gallon sample was
collected from R.B. White's iron phosphating/degreasing bath and shipped to HWRIC by a DOT licensed
hazardous materials transporter.   This solution was filtered  through the  UF pilot-scale unit, and
transmembrane pressure, concentrate flow, permeate flux, and influent temperature were monitored
during the test.
       In both pilot-scale tests, one-liter samples of permeate  and inflow were collected in glass
                                             25

-------
bottles at 1-hour intervals.  Permeate flux and temperature were measured every 15 minutes.  Samples
of the concentrate,  permeate,  and  unfiltered process solution were  analyzed for oil/grease  and
surfactant by TOC (Standard Method 5310 B) and for phosphate by titration (Standard Method 2310
B).

Results and Discussion
        Results of the preliminary pilot test with clean process solution  helped to identify the extent
of selective permeation of raw materials through the membrane.  Since the process solution was clean
and uncontaminated by oil,  Dura-Gard surfactants accounted for the only TOC contribution to the
system. Figure 9 shows that during the course of the test, surfactant TOC levels steadily rose in the
process tank while levels in the permeate showed a predominantly downward trend with the exception
of the last sampling point. Without emulsified oil in the process solution, all the surfactants were free
and able to interact with the membrane structure. The build up of free surfactant TOC in the feed tank
indicated that the PVDF membrane was selectively filtering out some free surfactant from the process
solution. As less and less surfactant permeated through the membrane, it appeared that the surfactant
molecules  were adsorbing to the membrane surface and forming micelles  in the membrane pores.
Interactions between free surfactant and membrane structures were described by Bhattacharyya, et.
al. and are summarized in the Waste Reduction section of this report.  It is interesting to note,
however, the sudden increase  in TOC/surfactant for  the final permeate  data  point.   During the
 preliminary pilot  test, the temperature of the feed stream rose from 25 to 51 °C (77 to  124°F) and
 concentration nearly doubled from 490 to 950 mg/L TOC.  As predicted by Bhattacharyya (1979),
 adsorption of free surfactant on the membrane was reduced as temperature and concentration
 increased.  For the full-scale testing, this was encouraging, and an increase in surfactant permeation
 was expected due to the warmer operating  temperatures (app. 140°F)  at R.B. White. As illustrated
 in Figure 10, phosphate levels in the process tank and the permeate remained relatively constant during
 the course of the test. Phosphate values for permeate were only slightly lower than in the process
 tank, indicating that the phosphating agents were not selectively removed by the PVDF membrane to
 a significant degree.  On the other hand, the surfactant component did show signs of selective removal
 by the PVDF membrane.
        Results of the second pilot-scale test conducted on  the 50-gallon sample of waste process
 solution are presented in Figures 11 and 12.  Figure 11a depicts the steady increase of TOC in the
 process tank and permeate with time.  As TOC levels built up in the tank,  more and more clean
 permeate  was filtered out and the oily process  solution  concentrated  down.   During the test
 temperatures rose from 24 to 54°C (75 to 130°F) and paralleled the rise in permeate TOC (see Figure
  11 b). Exactly what percentage of the TOC could be attributed to oil/grease or surfactant was unclear,
                                              26

-------
                                        •it- *
I

w
O)



1
Q.
U
o
       1000
        900-
        800-
        700 -
        600-
        500
                                                  190
        400
                                                  120
           0.0
                                                2.0
                                                         en
                                                         E
                                                         (B
                                                         Q>

                                                         £


                                                         I
                                                         O
                                                               TOC in Process Tank

                                                               TOC in Permeate
                    0.5       1.0       1.5

                    Elapsed Time (hours)

Figure  9:  TOC  in Process  Tank and  Permeate  vs Time

               Using  Clean Process  Solution
e

'a

a.
to
o
JK
Q.
       15200




       14800




       14400




       14000




       13600




       13200




       12800-
      12400
                                                               •a	  Process Tank

                                                               -«	  Permeate
           0.0
                                                          2.0
                      0.5         1.0          1.5

                         Elapsed  Time  (hours)

Figure 10:   Phosphate in Process Tank and  Permeate  vs Time

                Using Clean Process Solution
                                   27

-------
 D)
 E
 M
 O
 P
o
     4000
     3000-
2000-
     1000
                                         100
                                        -90
                                            -80
                                            -70
                                            r60
                                             50
           2468
            Elapsed Time (hours)
                                           10
                                                        TOG in Process Tank
                                                        TOC in Permeate
O)
^
o
"5
i
                                               u
                                               o
Figure  11a:
         TOC  in  Process Tank  and  Permeate vs  Time
            Using Waste Process Solution
                                              100
  o
  O)
  c
  •o
  2
  o
  a.
  c
      30-
      20
                                                          Temperature
                                                          TOC in Permeate
         02468
                 Elapsed Time (hours)

  Figure  11 b:   Temperature Effect  on Permeate  TOC vs Time
                   Using Waste Process Solution
                               28

-------
0>
E
4>
«*
CO
JZ
o.

o

Q.
5800




5600-




5400-




5200-




5000 -




4800-




4600-




4400-



4200
                                      Phosphate in Process Tank

                                      Phosphate in Permeate
                               4        6

                          Elapsed Time (hours)
                                                  10
 Figure 12:   Phosphate in  Process Tank  and  Permeate  vs Time
                 Using Waste Process Solution
                                     29

-------
but research by Bhattacharyya and results of the first preliminary pilot test led to the assumption that
perhaps increasing levels of surfactant TOC were passing in the permeate along with some oil.
       As demonstrated by the two pilot-scale tests, the interactions of free and combined surfactant
with the membrane surface  are completely different.  The preliminary pilot test with clean process
solution  demonstrated that  free  surfactant tended  to foul the membrane,  increasing  surfactant
rejection. When temperature and concentration in the feed tank  increased,  surfactant  adsorption
occurred to a lesser extent.  In the second pilot-scale test with the waste process solution, more
surfactant was tied  up with the  emulsified oil droplets and was not available to adsorb onto the
membrane surface. Only a small fraction of free surfactant was actually present in the waste process
solution; most of it became physically associated with the oil during the rigorous mixing in the process
tank. With the waste process solution, less fouling was attributed to adsorption  of surfactants and
formation of micelles on the membrane,  and an increase in temperature seemed to  encourage
permeation of surfactant.
        Figure 12 illustrates the  relationship between  phosphate levels in the  process tank and
corresponding levels in the permeate in the second pilot-scale test with the waste process solution.
The graphs show an upward trend in phosphate levels for both concentrate and permeate with time.
Values for both data sets were relatively close, confirming the free passage of phosphating chemicals
through the PVDF membrane. The increasing levels of phosphate with time were not anticipated in
the preliminary  pilot test with the clean  process solution.   This may be an indication that the
phosphating agents  were somehow associated  with the emulsified oil or other rejected components
in the feed tank and passed through as concentrations rose in the feed tank.
        Results from both pilot tests indicated that UF would be effective in removing emulsified oil
contamination while conserving some valuable raw materials. Data confirmed that phosphating  agents
were conserved during ultrafiltration. Levels in the permeate were only slightly less than the levels in
the feed stream.  The interactions of free and combined  surfactants with the  membrane surface
indicated that some free surfactant was selectively removed by the PVDF membrane, but the presence
 of oil in the full-scale application  was expected to minimize fouling caused by surfactant adsorption
 and micelle formation within the membrane structure.
                                              30

-------
                                          Section 6
                         FULL-SCALE IMPLEMENTATION AND TESTING
Methods and Materials
       The full-scale in-plant testing featured an ultrafiltration system provided by Koch Membrane
Systems (Model UF-4) equipped with 4 of the same 1" tubular PVDF membranes {100,000 MWCO,
4.4 sq. ft. total area) utilized in the pilot phase.  The modified batch scheme was chosen for the full-
scale in-plant test (see Fig. 1, Sect. 1 and Fig. 8, Sect, 4).

Full-Scale Activities
       Results from the pilot study were used to develop a full-scale, modified-batch test conducted
onsite  at R.B. White's facility-  The full-scale test applied ultrafiltration directly to the 5000 gallon iron
phosphating/degreasing bath. The objective was to directly measure the effect of ultrafiltration on the
process solution under actual plant conditions. The full-scale test took into account the constant input
of oil from the production line and the daily addition of bath chemicals. Additionally, the full-scale test
also helped identify and develop solutions to problems with the ultrafiltration equipment and anticipate
operating changes that should be made on a permanent unit.
       Transmembrane pressure, flux, and temperature were monitored during the test as samples
were taken. A daily log sheet was utilized to keep records of chemical additions to the bath. Samples
of inflow (from the intermediate process tank) and permeate were collected daily in 1 -liter glass bottles
and stored in a 4°C cold room.  Samples of the bath were  collected every third day from five different
locations 1 foot deep along the length of the tank using a transfer pump. All samples were collected
in the  morning prior to production start-up. This helped to nullify any fluctuations in process solution
quality that occurred during the day under actual production conditions. Rinse water samples were
also   collected   every third   day from  the  catch basin  located  downstream  of  the  iron
phosphating/degreasing tank. These samples helped to determine whether oil was clinging to the parts
as they emerged from the bath and if UF would help  to  control compliance problems with  the
discharge.  All samples from the full-scale study were analyzed for oil/grease and surfactant by TOC
and for phosphate by titration (see Quality Assurance section). A representative group of 20 full-scale
samples were  analyzed  for  orthophosphate  by 1C and for oil/grease  and surfactant  by  the
extraction/infrared technique (Standard Method  5520C).  A sample  of the make-up water supply was
also tested for Ca/Mg hardness to identify  hardness levels that might contribute to higher surfactant
usage.
        To determine the impact of the UF system on product quality, DuBois Chemicals ran a series
of 6 steel test plates through the iron phosphating/degreasing bath during the UF experiment. The test
                                              31

-------
plates were  analyzed for coating weight,  rust creepage, paint thickness, and paint adhesion as
described in the Quality Assurance section.
        Permeate flux rates were recorded daily, and the time required for the UF unit to filter the entire
5000 gallons of process solution was calculated. When it had been determined that the UF unit had
processed 5000 gallons of permeate, a batch down of the contents in the intermediate process tank
was performed. The feed line from the 5000-gallon bath to the process tank was shut off, and the
UF unit concentrated down the 55-gallon process tank to approximately 10 gallons of oily waste.
        Data obtained from the full-scale modified batch test was used to determine whether UF would
be a viable option for waste reduction at the R.B. White plant.  Technical, operational, and economic
aspects associated with the UF equipment were examined  to evaluate the feasibility of this technology
to improve R.B. White's metal fabrication operation.

Results and Discussion
        Results of the full-scale testing on site at R.B. White provided a more complete picture of how
ultrafiltration impacted the 5000 gallon iron phosphating/degreasing bath and the quality of the parts
it produced. For the scope of this report, the time frame for experimental  data was limited  to the time
that was required to filter the first 5000 gallons of permeate and to proceed  to batch down. This
covered a test period of 11  days. A total of 180 UF operating hours was required to process 5000
gallons of bath solution.
        When  field testing began, the iron phosphating/degreasing bath had not been replaced in over
3 months. The aqueous solution was murky with dirt and oil, and large patches of free oil floated on
the surface. The changes that took place over the next 11 days of ultrafiltration testing  produced a
dramatic effect.  Surface oil slicks disappeared and were replaced by a clean, light foam. The bath
 solution was visibly clearer, and plant personnel testified  that it looked like a freshly recharged bath.
 Product quality tests also revealed that quality of the steel parts was improved significantly as a result
 of the cleaner process solution maintained by the ultrafiltration system.
        In order to quantify this exceptional improvement, results of the full-scale testing were analyzed
 one-step further than the bench- and pilot-scale tests,  paying particular attention to the separate TOC
 contributions  of oil and surfactant. Since standard analytical test methods were not adequate to make
 the  important distinction between oil/grease and surfactant for reasons detailed  in the  Quality
 Assurance section of this report,  analysis of the field data took a new approach. A mass balance was
 developed to  provide a more realistic representation of the events that took place over those 11 clays.
 The mass balance utilized TOC data,  records of chemical additions, and an estimation of the daily
 production  load.  The analytical  information on TQC provided a base  line for total oil/grease and
 surfactant levels in  the bath, process tank, permeate, and rinse water.  Daily records of the chemical
                                               32

-------
       500
.3-     400 -
"w
E
o
o

*»*
C3
"5
 3
 0)
       300-
       200 -
       100
             Best Fit:  y = 20.974 + 22.436X  RA2 = 0.999
           6      8      10     12    14     16     18     20   Dura-Gard  (g/L)

               1.0            1.5           2.0          2.5      Dura-Gard  (oz/gal)
Figure  13:   Calibration Curve  for  Dura-Gard  TOC
                                     33

-------
additions of Dura-Gard Soke provided an accurate measure for the amount of TOC added as surfactant
to the bath during the course of the full-scale test (see Figure 1 3 for calibration curve of Dura-Gard
TOC). An estimation of the daily production load provided a practical and realistic value for .the amount
of oil that was washed off in the bath during 8 full days of production.
       The daily production load was calculated by dividing the oil TOC in the bath at the start of the
test by the number of working days since the bath had last been dumped. The oil  TOC in the bath at
the start  of the test  was extrapolated from titration data that was used to determine Dura-Gard
concentrations in the  bath.  The titration was actually a measure of the level of phosphating agents
in the bath, but this method was also valid for tracking the level of surfactant before ultrafiltration
began.  According to DuBois, both the phosphating agents and surfactants are consumed at the same
rate under normal operating conditions. On the morning of the full-scale start-up,  the bath titrated at
0.80 ounces Dura-Gard per gallon. Using the titration data, the level of surfactant was computed from
the calibration curve in Figure 1 3 as  1 55 mg/L, and the level of oil was calculated as the difference
between total TOC (422 mg/L) and surfactant TOC (1 55 mg/L). This value of oil TOC (267 mg/L) was
converted to mass units (267 mg/L x  3.785 L/gal x 5000  gal) and divided by  71, the number of
working days since the bath had last been dumped. The estimated production load was computed as
0.07 12 kg oil/day.
        Once UF testing was implemented, the titration was no longer a reliable indicator of surfactant
 or oil levels. The selective rejection and permeation of surfactant and phosphating agents by the  UF
 membrane resulted in different consumption rates for each component, so the  mass balance was
 developed to characterize the levels of oil and surfactant in the  bath thereafter.
               MASS IN - CHANGE IN MASS =  MASS OUT                            (Eq. 2)
        where  MASS INn = (n(0) + m(S)}KW9d + {(O  + S)^ + (0 + S)^},,^
               CHANGE IN MASSn = {(0 + S)procew + (O + S),lut)fl0 + (0 + S)Mm}n
                                             tank
               MASS OUTn =  {(0 + S)^ + (0  +  S)rima}n
               n = number of  working days (n  = 0 to 8)                          .  ;
               m = number of Dura-Gard additions  (90 Ibs) in n days (m = 0 to  5)
               O = oil TOC mass
               S = surfactant  TOC mass
               Known Values:       nntS),,^       TOC added as Dura-Gard surfactant
               Measured Values:     {(O + S)tath> ^oc^ ,„*}„
               Estimated Values:     {(S),^,}^,,     Initial surfactant TOC mass
                                             o     Initial oil TOC mass
                                                  Daily production load
                                              34

-------
              Unknown Values:      {(O + S),lra8, ,lud(J)>, . Wm}n
                                    Ratio of 0 to S
       The known, measured, and estimated values of TOC mass account for the major contributions
to the mass balance equation.  The unknown values of {(O  + S)rirw>i auivti ,Wm}n and {(0  + S)rirwjn,0
made up a smaller fraction of the mass balance and were difficult to determine analytically.  The TOC
mass in the rinse was not calculated due to lack of consistent measurements for the volume of rinse
water; the  mass of sludge produced  was not possible to measure under field conditions; and the
skimmed oil did not lend itself to a homogeneous sample that could be analyzed by the HWRIC's TOC
analyzer. These unknown values of {(0 +  S)rlw. ,ludBe, .Wm}n and {{O + S)rlme}n=0 were combined into
a single loss factor, k, to complete the mass balance.  Still, one key assumption had to be made for
the unknown ratio of O to S. A quantitative relationship needed to be defined for the  ratio of oil/grease
to surfactant in  each facet of the  mass balance  equation: bath, process tank, and rinse.  Without
analytical laboratory data to affirm the relative concentration of each component in  the separate test
streams, insight into the physical chemistry of nonionic surfactants was used to fit together the final
pieces of the puzzle.
       Surfactant molecules associate with an oil phase in an amount proportional to the concentration
of oil present and the conditions of mixing.  Hundreds of surfactant molecules may  surround a single
droplet of oil to form a micelle, but the amount of surface area on the oil nucleus determines the
number of bonding sites for the surfactant's hydrophobic tails.  When sufficient surfactant is available,
the critical micelle concentration (cmc) is reached, and the surfactant molecules effectively emulsify
the oil in an  aqueous phase.  There is a direct and consistent correlation  between the amount of
surfactant and the  amount of oil  in  an emulsified solution (Cutler, Davis,  1972).  Of  course, the
presence of free surfactant or oil not associated with the emulsion would bias the  balance between
the oil to surfactant ratio.  Based on field observation during the full-scale testing, the only times free
oil was visible were on the first day in the bath and in the process tank there  after. At all other times,
the absence of free oil indicated that the balance of oil to surfactant favored excess surfactant. In
terms of the mass  balance, this observation led to a conservative estimate for reporting  the amount
of surfactant,  attributing a higher percentage of the total TOC to oil.
       The mass balance assumed that the  ratio of oil/grease to surfactant in each test stream (bath,
process tank, rinse) was equivalent to the ratio of oil to surfactant in the total system at time  n.  The
ratio  of 0 to S  was defined  using the daily production load, surfactant additions, and initial bath
conditions. An example calculation follows.
                                             35

-------
             MASS IN - CHANGE IN MASS = MASS OUT

      (n(0) + mtS)}.^ + {(O + S)taUl}n.0 - {(0 + S)procOM}n = {(O  + S)^}n  + k

      where k = {{0 + S),lud89 + (0 + S)tklm + (0 + S)rirtM}n - {(0  + S)rinM}n,0

      For    n = 8 days
             m = 5 additions
                    =  .0712 kg oil/day (daily production load)
                    =  -989 kg surfactant
                      o = 5.06 kg oil  TOC (initially)
             {(S)b«h}n=.o = 2-94 k9 surfactant TOC (initially)
             {(O + S)proc«}n.8 = 3.96 kg TOC
             {<0 + SltaJn^ =  6.43 kg TOC                                      |

      Substituting these values and solving for the loss factor, A:, gives:

             8(.0712)  +  5(.989)  + (5.06  + 2.94) -3.96 =  6.43  + k

             (.570 +  4.94)  + (5.06 + 2.94) - 3.96 - 6.43 = k = 3.12 kg TOC

      Grouping the O and S terms of the first two quantities separately and adding gives the total
      measurable TOC in the mass balance system at n = 8:

             OTOWI  = (-570 + 5-06) = 5-63 k0 oil TOC
             STOUI  =  (4.94 + 2.94) = 7.88 kg surfactant TOC

      Calculating the ratio of 0 to  S at n = 8 gives:

             O/S = 5.63/7.88  =  .714/1  or 0 = 41.7% and S = 58.3%

      Following the assumption that the ratio of oil/grease to surfactant is equivalent for each test
      stream (bath, rinse, process  tank, permeate) gives:
                   Jn-e = 41 .7% (340 mg/L TOC) = 142 mg/L oil TOC
              {(S)b.th}n-8 = 58.3% (340 mg/L TOC) = 198 mg/L surfactant TOC

              {(O)rlma}n»8 = 41 .7% (1 1  mg/L TOC) = 4.58 mg/L oil TOC
              {(S)rln«}n=8 = 58.3% (1 1 mg/L TOC) = 6.42 mg/L surfactant TOC

              {(0)proe<>tt}n.8 = 41.7%  (19000 mg/L TOC) = 7918 mg/L oil TOC
              {(S)pr0ee«}n»8  = 58.3% (1 9000 mg/L TOC) = 11082 mg/L surfactant TOC

              {(OUmMte}n,8 = 41 .7% (1 60 mg/L TOC)  = 66.7 mg/L oil TOC
                   m.atJn-B = 58.3% (160 mg/L TOC) = 93.28 mg/L surfactant TOC
       Figures 1 4a and 1 4b show the change in TOC levels of oil and surfactant in the bath and rinse
water over the course of the 1 1 -day experiment. When the field testing began, the bath had not been
replaced in over 3 months and free oil could be seen floating on the surface.  At the start of the* field


                                            36

-------
       400 -
  o»
  E
  O
  o
300-
       200-
       100
                                       Increased
                                       Dura-Gard
                                       additions
a  Total TOG
•  Surfactant TOC
•  Oil TOC
           0   20  40   60  80  100  120  140 160  180 200
                      Operating  Time (hours)

Figure  14a:   Oil and  Surfactant  TOC  in  Bath  vs Time
  o>
  E
                                                           0  Total TOC
                                                           •  Surfactant TOC
                                                           •  Oil TOC
        10-
         5-
           0   20   40  60   80  100 120  140  160  180  200

                       Operating Time  (hours)

Figure 14b:   Oil  and  Surfactant TOC  in  Rinse vs  Time

                                    37

-------
 TO
 E,

 O
     20000
     15000 -
 10000-
      5000-
                                                          Q  Total TOG
                                                          •  Surfactant TOO
                                                          a  Oil TOG
        -i—i—i—r
      0   20   40
-T
60
                           -i—•—i—•—i—•—i—•—i—•—r
                           80  100  120  140 160  180  200
                        Operating  Time  (hours)
Figure  14c:  Oil  and  Surfactant TOG  in Process  Tank  vs Time
        200
         150-
O
         100-
                                                            m  Total TOC
                                                            •  Surfactant TOC
                                                            •  Oil TOC
          -»—r
            20
                                      r—i—i—'—i—"—r
                    40   60  80  100  120 140  160  180 200
                           Operating Time

  Figure 14d:   Oil and Surfactant TOC  in  Permeate  vs Time

                                     38

-------
test, levels of oil/grease averaged 267 mg/LTOC and surfactant levels averaged 155 mg/LTOC. The
                                     :«   -•-•••»         -
bath titrated at 0.80 ounces Dura-Gard per gallon which was we!! below normal operating parameters.
Under normal conditions, the bath is maintained at a Dura-Gard concentration of 1.5 to 2 ounces per
gallon. On the third day, oil TOC in the bath had dropped by 157 mg/L to 109 rng/L. On this day, the
bath  composition  was still low  at 0.79  oz.  Dura-Gard/gal as measured by titration in  HWRIC's
laboratory.  Apparently, R.B. White's simple colormetric test  kit was not very precise in measuring
Dura-Gard levels in the bath.  HWRIC field personnel alerted plant operators and  provided R.B. White
with a more precise titration apparatus for testing the level of raw materials in the bath.  The new test
apparatus included a pipette, buret, and pH meter for more accurate sampling and pH measurements
than the simple test kit could provide. The same titration procedure used in the lab was also used in
the field and is described in the Quality Assurance section of this report.
       After ultrafiltration  testing began, R.B. White personnel began adding more Dura-Gard to the
bath, and phosphate levels and surfactant TOC concentrations increased substantially.  The rise in
phosphate levels in  the  bath shown in  Figure 15  confirmed this surge in raw chemical addition.
Phosphate levels in the bath, permeate, and process tank rose simultaneously with the increase in bath
concentrations.   The  close  correlation  between phosphate levels  in  the  separate  test streams
demonstrated that virtually all the phosphating agents were recovered by the ultrafiltration equipment.
However,  only a fraction of  the  unused surfactants managed to permeate through the  membrane.
Figures 14c and 14d reveal the discrepancy of surfactant TOC in the process tank and permeate. As
predicted by the two pilot-scale tests, ultrafiltration was preferentially removing surfactant.
        In addition to the effects  of ultrafiltration, increased surfactant consumption may have been
aggravated by high  calcium  and magnesium  levels in the make-up water supply.  Results of the
hardness analysis  on the make-up water identified 158 mg/L as CaCO3, which qualified the make-up
supply as a very hard water.  High levels of Ca and Mg precipitate soaps and surfactants like those in
R.B. White's bath and increase surfactant consumption although low pH levels help to minimize the
effect.  For R.B. White a water softening system might be beneficial, but the expense of purchasing
such a system would outweigh the cost of additional surfactant usage that R.B. White had already
become accustomed to before ultrafiltration was implemented.
        Since Dura-Gard Soke only contained 2-3% surfactant by weight, compensating for surfactant
lost to  ultrafiltration by adding more Dura-Gard could  have resulted in excessively high phosphate
concentrations that would have contributed to loose and dusty coatings.  At this point, it became clear
that Dura-Gard and Tart alone could not maintain the chemical  balance in the bath. DuBois Chemicals
recommended a neutral cleaning additive, IPI-27, that contained the surfactant component found in
Dura-Gard Soke along with very  low levels of phosphates.  Since there was no  good way to detect
levels of surfactant  in the bath without laboratory analysis,  it was suggested that plant operators
                                              39

-------
CO
0.
o
JC
CL
       7000
       6000-
       5000 •
       4000 n
       3000-
       2000
             Increase in Phosphate due to
             increased Dura-Gard additions
                                                               Q  Phosphate in BATH
                                                               •  Phosphate in INFLOW
                                                               B  Phosphate in PERMEATE
            0   20   40   60   80  100  120  140  160 180  200
                          Operating Time  (hours)
Figure  15:   Phosphate  Levels in  Bath,  Inflow,  and
                 Permeate vs Time
                                      40

-------
develop their own gage for adding the surfactant component, IPI-27, on an as-needed basis. Operators
could even add excess surfactant  as a safety factor td ensure clean part quality.  Koch Membrane
Systems representatives  indicated  that excess surfactant would not irreversibly harm the membrane,
and  DuBois Chemicals  representatives saw no  danger in excess  surfactant interfering  with  the
phosphating process.  The only loss seemed to be the extra money spent on adding too much
surfactant.  Relative to the cost of the other raw  materials, this added cost would be minimal.
        Results of the coating  weight, rust creepage, paint thickness, and paint adhesion tests
conducted by DuBois Research Laboratory and R.B. White personnel on samples of steel parts indicated
that acceptable product quality was achieved during the UF full-scale test. No previous product quality
records were kept on file,  but  conversation with DuBois laboratory personnel and R.B. White plant
operators confirmed that the product quality achieved during the UF full-scale test was good for R.B.
White's application for controlled indoor air environments.  Product quality test results appear in Table
3.

Table 3:  Product Quality Test Results
PRODUCT QUALITY TEST
Coating Weight
Rust Creepage 1 18"
Salt Fog 5%
Humidity 96%
Paint Thickness
Paint Adhesion
RESULTS (averages)
50.4 mg/ft2
Passing at 24, Failed at 48 hrs
Passing at 504 hrs
1 .6 mil
Good*
        * Only qualitative data available from R.B. White personnel.

        Permeate flux rates exhibited excellent performance during the full-scale operation. Rgure 16
 shows the change in permeate flux with time and the approach to steady-state.  According to Koch
 Membrane Systems representatives, R.B. White's acidic iron phosphating/degreasing bath produced
 much higher permeate flux rates than  alkaline  cleaning baths in similar  UF  applications.   The
 combination of acidic conditions  (pH =  3.5 to 4.5), high temperature  (140°F), relatively low oil
 concentrations  (267 mg/L in a dirty bath), and surfactant usage contributed to a  very favorable
 performance.  High temperatures functioned to lower water  viscosity and  reduce free  surfactant
 adsorption while encouraging higher flux rates.  The presence of surfactants  also helped to keep the
 membrane surface clean by tying up the oil and dirt in an emulsion.
                                              41

-------
 o
 o>
 x
 a
 a>
 E
 t«
 o
 a.
      40
      35-
 E    30 J
      25-
      20
Figure 16a:
                    50          100         150

                        Elapsed  Time  (hours)


                Permeate  Flux  Rate  vs  Time
                                                       200
      40
CO
o>
X


UL


»
       35-
       30-
       25-
       20
                     Temporary flux increase

                     due to cleaning filter trap


                         Batch-down performed and

                         process  tank  refilled with

                         more dilute process solution
                             -O-Q-B-
                 100     200     300    400

                        Elapsed  Time (hours)
                                                500
                                                         600
Figure  16b:   Permeate  Flux  Rate vs  Time  (Extended  Study)
                                  42

-------
       During the 180 hour operating period of the full-scale UF test, permeate flux rates remained
steady and the membranes were never cleaned. Data collection was completed, but the UF system
was  kept  on line to evaluate the equipment's long-term performance.  After  several weeks the
permeate flux began to decline rapidly.  Attempts to clean the membranes with alkaline cleaner (KLD-
Koch Liquid Detergent), diluted Tart (phosphoric acid plus surfactant), and rubber sponge balls were
                                                                     jet
not effective in restoring flux. These cleaners were designed to remove oil, but apparently the oil was
not the primary cause of fouling and flux decline. The other major component in the bath that had the
potential to foul the membranes was the iron phosphate precipitate.  A two-stage cleaning regime with
rubber sponge balls was prescribed.  Sodium bisulfite was used in the first stage to reduce the iron
species to its more soluble state. Citric acid in the second stage acted as a chelating agent to combine
with the iron and prevent it from redepositing on the membrane surface. After lime pH adjustment,
the spent sodium bisulfite and citric acid  cleaners were acceptable to pour down the drain.   This
cleaning regime provided excellent flux recovery, but the high flux  rates only lasted a few hours.
       About this time, the centrifugal pump on the full-scale UF system began to show accelerated
signs of wear and tear. At least one pipe plug had been completely eaten away, and the pump casing
was leaking badly.  The  pump  was made of  cast iron and did not fare well under the corrosive
conditions of the bath and cleaning chemicals. The pump was taken  off-line and replaced with a
stainless steel pump.  When the interior of the old cast iron pump was examined, the  inside had
severely corroded  and the impeller  had  almost completely  disintegrated.  The  bath and cleaning
chemicals had dissolved the iron from the pump and precipitated  an iron phosphate coating on the
membrane surface.  The accelerated iron corrosion and  deposition  from the cast  iron pump kept the
membranes  from staying clean for longer than a few hours. When the pump was replaced with a
stainless-steel one and the membranes were cleaned weekly, permeate flux rates were restored and
maintained.
        In addition to membrane life, another long term concern pertained to regular maintenance on
the 5000 gallon tank.  Since ultrafiltration had  been implemented, periodic emptying of the tank was
eliminated.  Previously, emptying the tank allowed plant personnel to recover lost parts from the tank
floor and remove the scale and sludge that had accumulated on the tank and heating element surfaces.
The scale and sludge consisted mainly of iron phosphate precipitate and undissolved Dura-Gard Soke
powder.  DuBois recommended  that the scale  and  sludge  be removed  periodically  to reduce
contamination and improve heating element efficiency. Without emptying the tank, R.B. White needed
an alternative means of performing regular tank maintenance.
        Plant personnel found that they could recover lost parts just by running a long pole with a hook
at one end along the  bottom of the tank and  fishing out the steel parts and hangers. The heating
elements  were removable and could be cleaned  without emptying the tank.  With slight design
                                            43

-------
modifications, the sludge that settled on the bottom of the tank could be removed using a slow-moving
sludge scraping mechanism and a sludge pump at a low point in the tank. Another alternative for
regular tank maintenance might involve renting or purchasing a 5000 gallon holding tank to use every
other year to store bath solution from the 5000 gallon process tank while regular tank maintenance
is performed. This way R.B. White could empty the tank, clean away sludge and lost parts, and still
save the bath solution. For the meantime, R.B. White has chosen to go with a third alternative.
DuBois Chemicals has recommended a liquid form of Dura-Gard Soke known as Secure Soke. This was
intended to reduce the amount of sludge accumulated in the tank by eliminating the undissolved Dura-
Gard component.
                                             44

-------
                                          Section 7
                                    QUALITY ASSURANCE
       A Quality Assurance Project Plan (QAPP) was prepared and approved by the U.S. EPA before
testing began (Miller, Lindsey, 1992). The QAPP contains a detailed design for conducting this study.
The experimental matrix,  sampling and testing  procedures, and laboratory analytical procedures are
delineated  in the QAPP.  The analytical  objectives of the QAPP are discussed below.  With  the
exception of analytical difficulties involved in the oil/grease analysis, QA objectives for this project
were not only met but expanded to include more complete sample characterization.
       The original intent, as outlined in the QAPP, was to analyze all samples for oil and grease in
order to determine  the proficiency of ultrafiltration to remove  oil contamination from  the iron
phosphating/degreasing bath.  As the project progressed, it became apparent that samples should also
be analyzed for concentrations of surfactant and phosphating agents to determine if  valuable raw
materials were being filtered out along with the oil.  Additionally, steel test panels were analyzed for
product quality during the full-scale study. Analysis of samples from the  bench-, pilot-,  and full-scale
studies  included  tests  for   oil/grease,   polyhydrocarbons  (PHC),  total  organic  carbon  (TOO,
orthophosphate, EDTA hardness, coating weight, rust creepage, paint thickness, and paint adhesion.
Table 4 presents a summary of sample types and analyses.
Table 4: Sample Summary
PHASE
Bench Scale
Pilot Scale
Full Scale
SAMPLE TYPE*
Process Solution
Concentrate
Permeate
Inflow
Permeate
Process Solution (Bath)
Inflow (Process Tank)
Permeate
Rinse Water
Raw Water Supply
Phosphated Steel Panels
Painted Steel Panels
SAMPLES
19
18
18
18
22
20
10
10
5
1
3
3
ANALYSIS
Oil/Grease
Oil/Grease
Oil/Grease
TOC, Phosphate
TOC, Phosphate
Oil/Grease, PHC, TOC, Phosphate
Oil/Grease, PHC, TOC, Phosphate
Oil/Grease, PHC, TOC, Phosphate
Oil/Grease, PHC, TOC, Phosphate
EDTA Hardness, TOC
Coating Weight
Rust Creepage, Thickness, Adhesion
        Liquid samples were stored at 4°C in glass bottles with Teflon-lined caps;
        holding times did not exceed 2 months.
                                              45

-------
       Laboratory analyses of the samples collected during this three-phase project were modified as
new information came to light.  Oil and grease analysis as prescribed by USEPA  Method 413.1
(Methods for Chemical Analysis of Water and Wastes,  1983) was complicated by the presence of
nonionic surfactants in the process solution. As the project progressed, it became evident that: the
nonionic surfactants in the process solution would interfere with the accuracy  of this analytical
procedure. The surfactants contributed to a persistent, stable emulsion layer in the separatory funnel
that inhibited complete solvent  extraction.  Attempts to break the  emulsion were time-intensive and
only partially successful.  Glass beads, glass wool, Na2S04, NaCI,  heating, freezing, pH adjustment
centrifugation,  and continuous solvent-solvent extraction were not effective in breaking the emulsion
or achieving complete extraction.
        In addition to the  problems associated with the emulsions, the surfactants also contributed
false readings  regarding the amount of oil and  grease  present.  No known solvent  will .selectively
dissolve only oil and grease. With the USEPA  partition-gravimetric method  (Method 413.1),  freon
extraction not only dissolved oil and grease but other organics as well (Clesceri, Greenberg,  Trussed,
1989). In this  case, biodegradable nonionic surfactants were extracted along with  the oil and grease
in the process solution.  When the extract was distilled, the residue reflected the combined;total mass
of oil/grease plus surfactant.   Since each  component carried a different meaning,  measuring the
quantity of the two components separately as oil/grease and surfactant was important.
        Standard Method 5520  C extraction/infrared technique (Standard Methods for the Examination
of Water and Wastewater, 1989) goes one step beyond the USEPA partition-gravimetric method to
differentiate between  oil/grease and other  organics.  This method utilizes solvent  extraction plus
infrared spectroscopy.  Analysis of the solvent  extract by infrared spectrophotometer is capable of
discerning oil and grease as polyhydrocarbons (PHC) apart from nonionic surfactants. A representative
group of 20 full-scale samples  were analyzed using this method, but the  same problems associated
with the extraction procedure kept these samples  from providing any conclusive data.
        During the course of this study, laboratory personnel found that standard methods, would not
be adequate for analyzing the industrial process solution. Following standard protocol for defining and
reporting levels of oil and grease provided false and misleading information. In-house laboratory staff
worked on new methods and combinations of methods that would give a more accurate indication of
the samples' make-up. Their research went beyond the scope of this pollution prevention project, but
 HWRIC staff will continue to research the analysis of similar industrial solutions in hopes of establishing
 a more complete and accurate  procedure for analyzing oily-detergent wastewaters.
         Due to the complications involved in the oil and grease analysis, another analytical  test was
 developed for the complete set  of pilot- and full-scale samples.  Analysis for total organic carbon (TOO
 was used as a surrogate test for combined levels of oil/grease and non-ionic surfactant.  Analysis by
                                              46

-------
TOC provided an  accurate, efficient, and  repeatable test for combined  oil/grease and surfactant
concentrations.   Under the circumstances? TOC  provldeS the  best  indicator of the ability of
ultrafiltration to remove organic species, whether oil or surfactant, from the process solution. A mass
balance was then developed to describe the fate of each component. Analysis for TOC was conducted
by the Rosemount Analytical Dohrmann DC-190 TOC Analyzer according to the Combustion-Infrared
Method (Standard Method 5310 B). The TOC Analyzer was calibrated with 1000 ppm potassium
hydrogen phthalate (KHP) giving a reference standard of 999.7 ppm C as KHP with a typical recovery
of 986.4 +/- 25.57, standard deviation  = 2.592 %. Method blank results were below the 10 ppm
detection limit.  Samples were diluted according to strength.
        In addition to  testing  for oil/grease and surfactant,  samples were analyzed for phosphate
concentrations to determine whether ultrafiltration might selectively remove the valuable phosphating
agents from solution along with the emulsified oil.  Concentrations of PO4 were determined by Ion
Chromatograph  (1C) analysis  using the Dionex  Ion  Chromatograph.  The  1C  was calibrated with
potassium  phosphate.  Samples were prefiltered with a solid phase extraction cartridge (C18) to
remove oil that might interfere with the 1C equipment. Due to the time and expense involved with the
1C testing, only  a representative group of 20 full-scale samples were analyzed for orthophosphate by
1C. The remaining samples were analyzed for PO4  by using the  same titration procedure R.B. White
plant personnel relied  on for  checking levels of phosphoric acid  (free  and combined) in the bath.
Samples were titrated  with .1  N NaOH to the second pK, endpoint of 9.2.  For the 20 representative
samples, a consistent  ratio was defined between the volume  of titrant required to reach the endpoint
and the known  P04 values determined by 1C. This ratio was used to calculate concentrations for all
remaining samples. P04 data  and data reduction procedures from the  titrations and 1C testing are
presented in the Appendix.
        For the onsite,  full-scale testing, the make-up water supply for the iron phosphating/degreasing
bath  was also tested for calcium/magnesium hardness. Hardness testing was used to determine the
potential for surfactant loss due to precipitation caused by calcium or magnesium in the water.  The
 EDTA titrimetric  method  (Standard Method 2340 C) was  used to determine the level of water
 hardness.  These results were utilized to determine  if a water softening system would be beneficial in
 conserving surfactant in the long run.
        Besides waste reduction, the primary objective of this study was to preserve product quality
 by ensuring the steel parts were  being cleaned and a good phosphate coating was being formed.
 Quality control tests were performed by the chemical supplier and R.B. White personnel while the full-
 scale ultrafiltration system was in operation. Standard tests  included coating weight, rust creepage,
 paint thickness,  and  paint adhesion.   Coating weight was  determined  using DuBois Research
 Laboratory's own method (Gortsas, 1993). Briefly described, steel test parts were processed through
                                              47

-------
the iron phosphating/degreasing bath during normal operation. The parts were dried, weighed, and
measured.  The phosphate coating was stripped with chromic  acid (Cr03), and the  parts were
reweighed.  The net difference in weight gave milligrams phosphate coating weight per square foot.
Rust creepage was determined according to ASTM methods B117 and D2247-87. For these tests,
steel test parts were processed through the complete production line.  Painted test parts were scored
and subjected to a either a 5% salt fog or 96% humidity. Rust creepage from the score was reported
as passing/failing for 1 /8" creepage at 24, 48, and 504 hours.  Paint thickness was measured by the
Zorelco 757 coating gage and calibrated with a standard film thickness. Paint adhesion was checked
by R.B. White plant personnel using the cross-hatch test which involved scoring the paint and applying
a piece of tape to the scratch.  Adhesion quality was observed as the tape  was pulled away, and
results were qualitatively reported.
       Tables 5 and 6  present the precision and accuracy data  for the full-scale sample analysis
performed by HWRIC and DuBois labs.  Due to the complications discussed earlier, data from the
oil/grease and PHC analyses was not used in the final evaluation.

Table 5:  Precision Data for Full-Scale Sample Analysis (Typical values)
PARAMETER
TOC
Oil/Grease"
PHC"
Orthophosphate
EDTA Hardness
Coating Weight
RUST CREEPAGE
Salt Fog 5%
Humidity 96%
Paint Thickness
NO.
2
13
REGULAR
SAMPLE
440 mg/L
42 mg/L
LABORATORY
DUPLICATE
430 mg/L
FIELD
DUPLICATE
39 mg/L
PRECISION
RPD = 2.3%
RPD>7.4%
Data from the oil/grease and PHC analyses were not utilized due to
complications in laboratory procedures.
46
14
100
114
111
113
112
6055 mg/L
3641 mg/L
158.1 44 mg/L
53.2 mg/ft2
1/8" at <48hr
1/8" at >504hr
1 .6 mil
6095 mg/L
158. 144 mg/L
N/A
N/A
N/A
N/A
3799 mg/L
N/A
5 1.6 mg/ft2
1/8" at <48hr
N/A
1 .4 mil
RPD = 0.7%
RPD = 4.2%
RPD=0%
RPD=3.0%
RPD=0%
N/A
RPD = 13%
         Oil and Grease analysis (USEPA Method 413.1) measures oil plus surfactant.
         PHC analysis (Standard Method 5520 C) differentiates between oil and surfactant.
                                              48

-------
Table 6:  Accuracy Data for Full-Scale Sample Analysis (Typical values)
PARAMETER
TOC
Oil/Grease*
PHC"
Orthophosphate
EDTA Hardness
Coating Weight
RUST CREEPAGE
Salt Fog 5%
Humidity 96%
Paint Thickness
ID
NO.
REGULAR
SAMPLE
MATRIX
SPIKED
SAMPLE
MATRIX
SPIKED
VALUE
ACCURACY
(%
RECOVERY)
No data was obtained for spiked samples because oil would not stay
emulsified or remain in solution.
Data from the oil/grease and PHC analyses were not utilized due to
complications in laboratory procedures.
3'
100"
114
111
113
112
3905 mg/L
158.1 44 mg/1
as CaC03
53.2 mg/ft2
1/8" at <48hr
1/8" at >504hr
1 .6 mil
1 1 54 mg/l
1000 mg/L
as CaC03
N/A
N/A
N/A
N/A
1018 mg/l
1 004 mg/L
as CaCO3
N/A
N/A
N/A
N/A
88.3%
100.4%
N/A
N/A
N/A
N/A
       Oil and Grease analysis (USEPA Method 413.1) measures oil plus surfactant.
       PHC analysis (Standard Method 5520 C) differentiates between oil and surfactant.
    •  Bath
    b  Raw water supply
                                            49

-------
                                          Section 8
                                    ECONOMIC ANALYSIS
       The costs and benefits associated with installing an ultrafiltration system at the R.B. White
facility were analyzed to determine the economic feasibility of this technology. Assumptions regarding
inflation rate, discount rate, federal tax rate, depreciation schedule, project life, and various operating
expenses were entered into a LOTUS spreadsheet program  (General Electric, 1987) which calculated
a number of economic indices.  The assumptions utilized in these calculations and their sources are
presented in Table 7.

Table 7: Assumptions for Economic Calculations                                     >
               (based on 1992 conditions)
ITEM
Inflation Rate
Discount Rate
Federal Tax Rate
Labor Rate
Power Costs
Depreciation Schedule
Project Life
Raw Materials Costs*
Down Time Costs"
Salvage Value
Waste Disposal Costs
FACTOR
4%
7.72%
34%
$15/hour
$.10/kw
7 years
20 years
$900/bath
$1000/bath
$0
$1 /gallon
SOURCE
Consumer price index
10 year treasury bill rate + 0.5%
General Electric, (1987) ;
R.B. White
Illinois Power i
General Electric, (1987)
Koch Membrane Systems, Inc.
R.B. White
R.B. White
Koch Membrane Systems, Inc.
R.B. White
         * Raw Materials Costs include phosphating and degreasing chemicals required to charge a fresh
          bath.
        ** Down Time Costs include costs associated with lost production incurred during bath cleaning
          activities.

        Table 8 provides estimates of the capital and operating expenses as well as the cost savings
 that could be anticipated as a result of investment in an ultrafiltration system.  These estimates assume
 that the phosphating/degreasing bath would have to be disposed of and recharged at a frequency of
 once every 2 years following installation of the ultrafiltration system. This contrasts with the schedule
 utilized prior to installation of the ultrafiltration system which involved disposing of and recharging the
                                               50

-------
             to c
             CO 4
                             s

                             O T
                             CO

                             Is
                         *•
                         8
                             5
                             (O
             o 01
             as
             88

             CO
             OJ
oil
£1

                 CO
             §

             ss

   s
             Sfc
                              s
             Rg
                                i
                                          3
          s
ENDITU
tlon Unit
EG ENDITURES
lean Membranes
icals
Ch
tenan
p
p Motor
               s 8 gl
                !•§.•§.§
                r£b?l
               ff
mption
x80%
(23
Disposal
ga./year
VI
aw
de
£ 'O
If,
III
~il
 & &
                                                 CO
                                                 (9
1 SO
ath
                           51

-------
bath at a frequency of at least 3 times per year. Additionally, it was assumed that the wasted bath
could be discharged directly to the sanitary district (following neutralization of its acidic conditions for
a minimal cost) as a result of ultrafiltration processing.  This is appropriate since the bath would no
longer contain significant quantities of emulsified oil which had prevented discharge of the spent bath
prior to utilization of ultrafiltration.
        As shown in Table 8, installation of an ultrafiltration system would require significant capital
expenditures (approximately $12,000) associated with purchasing  and installing the equipment.
Additionally, significant operating expenses associated with cleaning and replacing the membranes,
equipment maintenance, and power consumption would be incurred over the projected 20 year life of
the equipment.  However, substantial savings associated with reduced raw materials usage, reduced
plant down time, and  reduced waste disposal costs would repay the capital and operating expenses
several times over.
        Based on the estimated expenditures, savings, and economic conditions described above, the
financial analysis program (General Electric, 1987) was  utilized to calculate the economic indices
described in Table 9.  As shown, the economics associated with installing an ultrafiltration;system at
the R.B. White facility are very favorable.  It is estimated that the payback period associated with this
technology is only 6.9 months. The net present value and implied rate of return indices are $152,143
 and 178%, respectively. Therefore, investment in an ultrafiltration system at the R.B. White facility
 represents a very attractive economic alternative.

 Table 9:  Economic Summary
Capital Invested
Payback Period
Net Present Value (NPV)
Interest Rate of Return (IRR)
$12,000
6.9 months ,
$152,143
178%
                                               52

-------
                                           Section 9
                           CONCLUSIONS AND RECOMMENDATIONS

Conclusions
       The overall  evaluation of this pollution  prevention  project was based  on ultrafiltration
performance, product quality, and economics.  Results of the full-scale ultrafiltration study indicated
that the concentration of oil in the iron phosphating/degreasing bath was substantially reduced and
maintained at acceptable operating  levels.  Virtually all of the unused  phosphating agents  were
conserved although a portion of the unused surfactants was not. Permeate flux rates exhibited
excellent performance during the acidic (pH = 3.5), high temperature (140°F) operation and were high
enough to continuously process the constant input of oil from the production line.  The entire 5000
gallon bath was processed in 180 ultrafiltration operating hours. Coating weight, rust creepage, and
paint adhesion tests conducted by DuBois Research  Laboratory and R.B. White plant personnel on
samples of steel parts indicated that product quality achieved during the full-scale study was good for
R.B. White's application.  Based on  the  estimated expenditures  and  savings, the payback period
associated with this pollution prevention technology was only 6.9  months.
        By using ultrafiltration, the R.B. White will reduce its hazardous waste generation by at least
15,000 gallons  per year, a 99.8% reduction. In addition, the company will no longer have periodic
problems exceeding allowed oil and grease levels in their rinse water discharge to the sanitary district.
        This  project has successfully demonstrated  the ability of membrane filtration  to  reduce
industrial waste generation and recover valuable raw materials in a metal fabricating operation.  This
application introduces another innovative waste reduction technique to the metal fabricated products
industry that could benefit the many plants nationwide that use aqueous cleaner systems like the iron
phosphating/degreasing  process at the R.B. White facility.  The ultrafiltration system implemented  in
this project saves money, maintains good product quality, and reduces waste generation.

Recommendations
        Many industries are facing tougher laws and tighter restrictions on solvent use and hazardous
waste  generation.  Along  with the  environmental  concerns  come  the rising  costs  of liability,
compliance, and waste disposal.  Developing cleaner  production techniques will call upon innovative
ideas and creative engineering to build the environmentally responsible industry of the future.
        The metal fabricated products industry needs to examine its present methods of operation and
set goals for cleaner production.   Using  iron phosphate and alternative coating chemicals can
modernize a metal coating operation as existing and projected EPA regulations require more stringent

                                              53

-------
control and more expensive waste treatment methods for chromated seals, zinc  phosphate, and
chromate conversion coatings.  Making the switch from solvent to aqueous cleaners holds the potential
to reduce solvent waste, improve worker safety, and avoid stricter environmental regulations. Learning
to extend  the  life  of  aqueous process  solutions opens the door  to recovery technologies like
ultrafiltration that could save money in raw materials and waste disposal.  In-plant evaluation and
testing can provide useful information for customizing a recovery scheme to fit an industry's specific
needs. Developing and implementing new ideas can bring about innovative solutions, but the industry
should be aware of limitations and special concerns. The issues of membrane fouling, sludge build-up
and disposal, regular tank maintenance, proper chemical balance, and raw material substitution need
to be dealt with on a case-by-case basis.
       When industry  cooperates and  works with the EPA and equipment dealers, environmentally
responsible solutions to generating hazardous and industrial process wastes can be found. With the
advent of water-based cleaners, waste minimization and pollution prevention have become realistic and
attainable goals for the metal fabricated products industry.
                                              54

-------
REFERENCES

Bailey, P.A. The Treatment of Waste Emulsified Oils by Ultrafiltration:  Proceedings of the Filtration
Society.  Filtration and Separation. January/February 1977, pp. 53-55.

Bhattacharyya,  D., A.B. Jumawan, and R.B. Grieves.  Ultrafiltration Characteristics of Oil-Detergent-
Water Systems: Membrane Fouling Mechanisms. Separation Science and Technology. Vol. 14, No.
6, pp. 529-549,  1979.

Cheng, S.C., B.O. Desai, C.S. Smith, H.D. Troy, T.E. Ctvrtnicek, W.H. Hedely. Alternative Treatment
of Organic Solvents and Sludges from Metal Finishing Operations.  Cincinnati, Ohio:  U.S.  EPA.
September 1983. EPA/600/2-83/094.  PB84-102151.

Cheryan, Munir. Ultrafiltration Handbook.  Lancaster, Pennsylvania:  Technomic  Publishing Co., Inc.
1986.

Clesceri, Lenore S., Arnold E. Greenberg, R. Rhodes Trussed.  Standard Methods  for the Examination
of Water and Wastewater.  Baltimore, Maryland:  Port City Press.  Seventeenth Edition, 1989.

Cutler, W.G. and R.C. Davis. Detergencv: Theory and Test Methods.  Surfactant Science Series: Vol.
5.  New York:  Marcel Dekker, Inc. 1972.

Donnelly, R.G., R.L. Goldsmith, K.J. McNulty, D.C. Grant, M. Tan. Treatment of Electroplating Wastes
by Reverse Osmosis.  Cincinnati, Ohio:  U.S. EPA. September 1976. EPA 600/2-76-261.

General Electric. Financial Analysis of Waste Management Alternatives. Fairfield, Connecticut. 1987.

Gortsas, Louis A.  Phosphate Coating Weights.  DuBois Research Laboratory, Sharonville, Ohio 45241.
1993.

Grinn, M.E., E.L.  Brown, and J.C. Harris.  Journal of the  American  Oil  Chemistry Society. Vol. 38,
1961.

Hindin, D.A., W.M. Burch, and D.L. Fort.  Pollution  Prevention Options in Metal Fabricated Products
Industries: A Bibliographic Report. Washington, D.C.: U.S. EPA. January 1992.EPA/560/8-92/001 A.

Laine, Jean-Michel, James P. Hagstrom, Mark M. Clark, and Joel Mallevialle. Effects of Ultrafiltation
on Membrane Composition. Journal of the American Water Works Association.  November 1989. pp.
61-67.

Lee,  Soobok, Yves Aurelle, and Henry Roques.   Concentration Polarization, Membrane Fouling and
Cleaning in Ultrafiltration of Soluble Oil. Journal of  Membrane Science.  Vol. 19,  1984, pp. 23-38.

Mehta, Suresh and Thomas Besore.  Alternative to Organic Solvents in Metal-Cleaning  Operations.
Savoy, Illinois: Hazardous Waste Research and Information Center.  July 1989.

Miller, G.D. and T.C. Lindsey. Quality Assurance Project  Plan for Hazardous Waste Reduction  for a
Commercial Iron  Phosphating/Degreasing  Bath.  Cincinnati,  Ohio:  U.S. Environmental Protection
Agency.  February 1992.

Morrison,  Robert and Robert Boyd.  Organic Chemistry.  Third  Edition.  Boston, Massachusetts:
Allynand and Bacon, Inc.  1973. pp. 1059-1062.

                                             55

-------
Paul, Swaraj. Surface Coatings: Science and Technology. Chichester: John Wiley and Sons, 1986.

Pinto, Steven D.  Ultrafiltration for Dewatering of Waste Emulsified Oils.  Lubrication Challenges in
Metalworking and Processing: Proceedings, First International Conference.  IIT Research Institute,
Chicago, Illinois. June 7-9, 1978.

Phillips, Dick.  Practical Application of the Principles Governing the Iron Phosphate Process. Plating
and Surface Finishing, Vol. 77, No. 3, March 1990, pp. 31-35.

Springborn Laboratories, Inc.  Emulsified Industrial Oils Recycling.  U.S. Department of Energy. April
1982. DOE/BC/10183-1.

Quitmeyer, JoAnn. Aqueous Cleaners Challenge Chlorinated Solvents.  Pollution Engineering. Vol. 23,
No. 13, December 1991, pp. 88-91.

Ultrafiltration Saves Oil, Cleans Waste.  Iron and Steel International, p. 391,  December 1979.

Viego, Armando.  Ultrafiltration System Saves $200,000 per Year in Disposal Costs. Plant Services.
July 1980, pp. 26-27.

Wahl, James R. and Lawrence S. Gould. How to Extend the Life of Washer Baths. Industrial Finishing.
September 1979, p. 47.

Wittke, William J.  Phosphate Coatings.  Metal Finishing Guidebook Directory.  1987.    ;

Wray, Thomas K.  Soaps and Detergents. Hazmat World. Vol. 5, No. 8, 1992.  pp. 82-83.

U.S. EPA. Guides to Pollution Prevention: The Fabricated Metal Products Industry. Washington, D.C.:
U.S. EPA.  July 1990. EPA/625/7-90/006.

U.S. EPA.  Methods for Chemical Analysis of Water and Wastes. 1983.

U.S. EPA.  Waste Minimization in Metal Parts Cleaning. Washington, D.C.: U.S. EPA. August 1989.
EPA/530-SW-89-049.
                                             56

-------
GLOSSARY
(Cheryan,  1986)

Concentrate   : the portion of the feed solution that is rejected by the membrane

Flux          : the volumetric flow rate of permeate per cross-sectiona! area per time

Fouling        : detected as the decrease in permeate flux over time and is a result of the changes in
               the membrane structure or interactions between the components of the feed stream
               and the membrane surface

MWCO        : molecular weight cut-off,  the minimum weight of a typical molecule that would be
               rejected by the membrane

Permeate      : the portion of the feed  solution that passes through the membrane

Rejection       : describes the degree to  which ultrafiltration prevents components in the feed solution
               from passing through the membrane
                                             57

-------
APPENDIX Numerical Data from Figures in Text
Figure 5a:  Comparison of Wastewater Flux Rates During Bench-Scale Testing

       MEMBRANE MODEL         MASS FLUX
                                 (grams permeate/minute)

       M2 Teflon                  1.100
       (30,000 MWCO)

       SN04 Cellulose Acetate      1 .080
       (20,000 MWCO)

       PM30 Polysulfone           .704
       (2,000 MWCO)

       M3 Teflon                  .209
       (500 MWCO)

       PM 10 Polysulfone           .198
       (10,000)

 Rgure 5b: Comparison of Normalized Flux Rates During Bench-Scale Testing

       MEMBRANE MODEL                NORMALIZED FLUX
                                                     "t«cov«v'»'lnltltl
       M2 Teflon                 .423                .869
       (30,000 MWCO)

       SN04 Cellulose Acetate      .284                .921
       (20,000 MWCO)

       PM30 Polysulfone           .017                .001
       (2,000 MWCO)

       M3 Teflon                 .275                .886
       (500 MWCO)

       PM 10 Polysulfone           .009                .020
       (10,000)
                                          58

-------
Figure 9: TOC in Process Tank and Permate vs Time Using Clean Process Solution
      ELAPSED TIME      TOC (mg/l):
      (hours)       Permeate      Process Tank
      0.00
      0.33
      0.68
      1.02
      1.35
      1.90
170
180
160
140
130
170
490
520
550
620
720
950
Figure 10:  Phosphate in Process Tank and Permeate vs Time Using Clean Process Solution

      ELAPSED TIME      PHOSPHATE (mg/J):
      (hours)       Permeate    Process Tank
      0.00
      0.33
      0.68
      1.02
      1.35
      1.90
13376
13693
13456
13218
13851
14010
13693
13930
13851
13930
Figure 11 a: TOC in Process Tank and Permeate vs Time Using Waste Process Solution

      Permeate                  Process Tank
      ELAPSED
      TIME (hrs)

      0.00
      0.18
      0.52
      0.77
      1.02
      1.28
      1.53
      2.53
      3.53
      4.53
      5.32
      7.53
      8.53
TOC
(mg/l)

52
53
56
56
57
56
57
60
69
79
85
81
91
ELAPSED
TIME (hrs)
0.63
2.28
3.28
4.28
5.28
6.28
7.28
8.28
9.28
TOC
(mg/l)
790
810
870
980
1200
1400
1800
2300
3100
                                        59

-------
Rgure 11b: Temperature Effect on Permate TOC vs Time Using Waste Process Solution
      ELAPSED
      TIME (hrs)

      0.00
      0.18
      0.52
      0.77
      1.02
      1.28
      1.53
      2.53
      3.53
      4.53
      5.32
      7.53
      8.53
TOC
(mg/1)

52
53
56
56
57
56
57
60
69
79
85
81
91
TEMP
24.0
24.0
25.5
27.0
29.5
31.5
34.5
40.0
46.0
50.0
53.0
50.5
51.0
 Rgure 12: Phosphate in Process Tank and Permeate vs Time Using Waste Process Solution

       Permeate                         Process Tank
       ELAPSED
       TIME (hrs)

       0.77
       2.53
       3.53
       4.53
       5.32
       7.53
       8.53
 PHOSPHATE
 (mg/l)

 4828
 4749
 4947
 4947
 5026
 5184
 5184
       ELAPSED
       TIME (hrs)

       0.63
       2.28
       3.28
       4.28
       5.28
       7.28
       8.28
        0.0
        66.5
        132
        180
 422
 232
 340
 340
 155
 123
 200
 198
      PHOSPHATE
      (mg/l)

      4986
      4986
      5144
      5144
      5224
      5382
      5382
 Rgure 14a:  Oil and Surfactant TOC in Bath vs Time

                                 TOC (mg/l):
       OPERATING
       TIME (hrs)    Total         Surfactant    Oil
267
109
140
142
                                          60

-------
Figure 14b: Oil and Surfactant TOC in Rinse vs Time

                                 TOC (mg/1):
       OPERATING
       TIME(hrs)     Total         Surfactant    Oil

       0.0          21           8            13
       66.5         844
       132          11           7            4
       180          11           6            5

Figure 14c: Oil and Surfactant TOC in Process Tank vs Time

                                 TOC (mg/l):
       OPERATING
       TIME (hrs)     Total         Surfactant    Oil
0.5
18.5
42.5
66.5
85.0
108.0
132.0
158.0
180.0
710
4700
6300
7100
5900
8000
12000
1600
19000
261
2301
3351
3752
3325
4482
7078
9387
11088
449
2399
2949
3348
2575
3518
4922
6613
7912
Figure 14d: Oil and Surfactant TOC in Permeate vs Time

                                 TOC (mg/l):
       OPERATING
       TIME (hrs)    Total         Surfactant    Oil
0.5
18.5
42.5
66.5
108.0
132.0
158
180
67
42
39
42
75
150
170
160
25
21
21
22
42
88-
100
93
42
21
18
20
33
62
70
67
                                          61

-------
Rgure 15: Phosphate Levels in Bath, Inflow, and Permeate vs Time

                               PHOSPHATE Img/l):
      OPERATING
      TIME (hrs)    Bath         Inflow       Permeate

      0.0         3617        3958
      185                     3245        3166
      42*5                     3007        2849
      66'.5        3562        3799        3561
      85.0                     3483        3324
      108.0                    4353        4749
      1320       6142        6094        6332
      158.0                    6094        6253
      180.0       6008        6174        6055

 Rgure 16: Permeate Flux Rate vs Time
ELAPSED
TIME (hrs)
18.5
42.5
66.5
108.0
132.0
158.0
180.0
252.0
276.0
300.0
324.0
348.0
516.0
FLUX
(gal/hr)
39.1
32.1
31.0
31.0
29.0
26.1
27.3
27.3
27.3
27.3
27.3
26.5
25.0
                                         62

-------
Data Correlation for Phosphate Analysis by Titration and 1C

      SAMPLE      TYPE
Volume       PO4 (mg/1)
Titrant (mL)   by 1C
RATIO
      3
      6
      7
      17
      21
      22
      30
      34
      33
      42
      43
      46
      53
      54
      55
      67
      70
      71
                                              AVERAGE     782.1

       (Volume of Titrant, ml) x {Ratio = 782.1, mg/l/ml) = P04 mg/1
Bath
Permeate
Inflow
Bath
Permeate
Inflow
Bath
Permeate
Inflow
Inflow
Permeate
Bath
Bath
Inflow
Permeate
Bath
Permeate
Inflow
4.4
5.2
5.0
4.5
4.6
4.8
7.9
8.0
7.7
7.8
7.6
7.6
7.8
7.8
7.6
9.7
9.4
9.3
3937
4091
4028
3455
3642
3736
6333
6161
5129
5588
6248
6357
5760
5428
6145
8011
7790
6909
894.8
786.7
805.6
767.8
791.7
778.3
801.6
770.1
666.1
716.4
822.1
836.4
738.5
695.9
808.6
825.9
828.7
742.9
                                          63

-------

-------