WATER POLLUTION CONTROL RESEARCH SERIES 12060 DXF 07/71
Membrane Processing of
Cottage Cheese Whey
for Pollution Abatement
U.S. ENVIRONMENTAL PROTECTION AGENCY
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes
the results and progress in the control and abatement
of pollution in our Nation's waters. They provide a
central source of information on the research, develop-
ment, and demonstration activities in the Environmental
Protection Agency, through inhouse research and grants
and contracts with Federal, State, and local agencies,
research institutions, and industrial organizations.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed to the Chief, Publications
Branch, Research Information Division, Research and
Monitoring, Environmental Protection Agency, Washington,
D. C. 20460.
-------
Membrane Processing of Cottage Cheese Whey
for Pollution Abatement
by
Crowley's Milk Company
Binghamton, New York
for the
OFFICE OF RESEARCH & MONITORING
ENVIRONMENTAL PROTECTION AGENCY
Project #12060 DXF
July 1971
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 • Price $1.25
-------
EPA Review Notice
This report has been reviewed by the Environ-
mental Protection Agency and approved for
publication. Approval does not signify that
the contents necessarily reflect the views
and policies of the Environmental Protection
Agency, nor does mention of trade names or
commerical products constitute endorsement
or recommendation for use.
-------
ABSTRACT
A two-step membrane process has been demonstrated for the
treatment of cottage cheese whey. The process produces val-
uable protein and lactose by-products while reducing the BOD
of the whey effluent. The process has been studied in de-
tail in prototype experiments at Abcor, Inc., Cambridge,
Mass., and in a 10,000 Ibs/day pilot plant at Crowley's Milk
Company, Binghamton, New York.
In the two-step process, a protein concentrate is first re-
covered in an ultrafiltration operation. In the second step,
ultrafiltration permeate (de-proteinized whey) is concentra-
ted by reverse osmosis to provide a lactose concentrate.
The protein concentrate can be further concentrated and/or
dried; and the lactose concentrate can be further concentra-
ted and lactose recovered by crystallization, or otherwise
processed. Alternatively, the membrane concentrates can be
used directly as fluid products.
Operation of the pilot plant was successful and almost trou-
blefree. BOD reduction of the raw whey was about 97%, from
an initial value of about 35,000 mg/1 to less than 1,000
mg/1. Membrane life was excellent, and membrane fluxes were
economically high.
The pilot plant produced protein and lactose products with
low total plate counts, and nil coliform counts. Using the
cleaning procedure developed in the prototype program, total
plate counts were typically below 50,000 org/ml.
Projected capital cost for a 300,000 Ibs cottage cheese whey/
day demonstration plant is $697,000. This includes both the
ultrafiltration and reverse osmosis sections of the plant,
tanks, and a building to house the plant. Projected opera-
ting costs are $220,000 per year. Projected income from
utilization of the protein and lactose concentrates makes
the process profitable and results in an attractive return
on investment.
This report was submitted in fulfillment of the requirements
of Grant No. 12060 DXF given by the Environmental Protectioi
Agency to Crowley's Milk Company-
111
-------
TABLE OF CONTENTS
Page
I. CONCLUSIONS 1
II. RECOMMENDATION 3
III. INTRODUCTION 5
BACKGROUND AND NEED FOR PROJECT 5
PRODUCTION OF COTTAGE CHEESE WHEY 5
THE COTTAGE CHEESE WHEY DISPOSAL
PROBLEM 7
PROPOSED MEMBRANE PROCESS 7
IV. PROTOTYPE EXPERIMENTS 15
ULTRAFILTRATION 15
EXPERIMENTAL PROCEDURE , 15
RESULTS 18
REVERSE OSMOSIS 37
EXPERIMENTAL PROCEDURE AND RESULTS 37
V. DESCRIPTION OF PILOT PLANT 57
ULTRAFILTRATION SECTION (LOW PRESSURE) 57
REVERSE OSMOSIS SECTION (HIGH PRESSURE) 60
VI. PILOT PLANT OPERATION 71
GENERAL PERFORMANCE 71
SAMPLING AND ANALYTICAL PROCEDURES 71
ULTRAFILTRATION SECTION 73
REVERSE OSMOSIS SECTION 81
MICROBIOLOGICAL DATA 99
v
-------
TABLE OF CONTENTS (CONT.)
Paqe
VII. DEMONSTRATION PLANT DESIGN AND PROCESS
ECONOMICS 105
PRELIMINARY DESIGN BASIS FOR AND
DESCRIPTION OF 300,000 LBS/DAY PLANT 105
CAPITAL COSTS 109
OPERATING COSTS 112
VIII. INDUSTRY INTEREST (INCLUDING PUBLICATIONS) 115
IX. ACKNOWLEDGEMENTS 121
X REFERENCES 123
Vl
-------
FIGURES
Page
1. TRENDS IN FLUID MILK PRODUCTION 8
2. OSMOSIS AND REVERSE OSMOSIS 10
3. MECHANISM OF MEMBRANE SEPARATION 11
4. MEMBRANE PROCESS FOR WHEY TREATMENT -
FLOW SCHEMATIC 13
5. TEST SYSTEM FLOW SCHEMATIC 16
6. ULTRAFILTRATION DATA: PROTEIN RETENTION 21
7. ULTRAFILTRATION DATA: GEL PERMEATION
CHROMATOGRAPHY ANALYSES 22
8. COMPOSITION OF WHEY PROTEIN CONCENTRATE 24
9. ULTRAFILTRATION DATA: DEPENDENCE OF FLUX
ON TIME 26
10. ULTRAFILTRATION DATA: DEPENDENCE OF FLUX
ON CIRCULATION RATE 28
11. ULTRAFILTRATION DATA: DEPENDENCE OF FLUX
ON PRESSURE 29
12. ULTRAFILTRATION DATA: DEPENDENCE OF FLUX
ON TEMPERATURE 30
13. ULTRAFILTRATION DATA: DEPENDENCE OF FLUX
ON PROTEIN CONCENTRATION 32
14. ULTRAFILTRATION DATA: FLUX AND PROTEIN
RETENTION 33
15. ULTRAFILTRATION DATA: BACTERIAL GROWTH
DURING WHEY CONCENTRATION 36
16. REVERSE OSMOSIS DATA: PERFORMANCE OF
DIFFERENT MODULES 39
17. REVERSE OSMOSIS DATA: COD REJECTION 40
18. REVERSE OSMOSIS DATA: LACTOSE REJECTION 41
vii
-------
FIGURES (CONT.)
Page
19. REVERSE OSMOSIS DATA: LACTIC ACID REJECTION 42
20. REVERSE OSMOSIS DATA: DEPENDENCE OF SOLIDS
REJECTION ON PRESSURE AND CONCENTRATION 44
21. REVERSE OSMOSIS DATA: DEPENDENCE OF COD
REJECTION ON PRESSURE AND CONCENTRATION 45
22. REVERSE OSMOSIS DATA: DEPENDENCE OF LACTOSE
REJECTION ON PRESSURE AND CONCENTRATION 46
23. REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
FEED SOLIDS CONCENTRATION 48
24. REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
FEED SOLIDS CONCENTRATION 49
25. REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
FEED SOLIDS CONCENTRATION 50
26. REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
CONCENTRATION AND PRESSURE 51
27- PILOT PLANT PHOTOGRAPH 58
28. PILOT PLANT PHOTOGRAPH 58
29. ULTRAFILTRATION SECTION 59
30. CIP SYSTEM 62
31. PILOT PLANT PHOTOGRAPH 65
32. HIGH PRESSURE PIPING DIAGRAM 66
33. PILOT PLANT PHOTOGRAPH 65
34. PILOT PLANT PHOTOGRAPH 69
35. PILOT PLANT PHOTOGRAPH 69
36. RAW WHEY ACIDITY 74
37. PILOT PLANT ULTRAFILTRATION DATA: EFFECT OF
FEED CONCENTRATION ON FLUX "* 76
Vlll
-------
FIGURES (CONT.)
Page
38. PILOT PLANT ULTRAFILTRATION DATA: WHEY
FLUX VS. TIME 77
39. PILOT PLANT ULTRAFILTRATION DATA: WHEY
FLUX VS. TIME 79
40. PILOT PLANT ULTRAFILTRATION DATA: PROTEIN
RETENTION 80
41. COD-CONDUCTIVITY CORRELATION FOR FEED AND
PERMEATE OF REVERSE OSMOSIS SYSTEM 83
42. PILOT PLANT REVERSE OSMOSIS DATA 84
43. PILOT PLANT REVERSE OSMOSIS DATA: DEPENDENCE
OF FLUX ON CONCENTRATION 85
44. PILOT PLANT REVERSE OSMOSIS DATA: DEPENDENCE
OF FLUX ON FEED CONCENTRATION 86
45. PILOT PLANT REVERSE OSMOSIS DATA: LACTIC
ACID REJECTION 87
46. PILOT PLANT REVERSE OSMOSIS DATA: STAGE 1
FLUX 89
47. PILOT PLANT REVERSE OSMOSIS DATA: STAGE 2
FLUX 90
48. PILOT PLANT REVERSE OSMOSIS DATA: STAGE 3
FLUX 91
49. PILOT PLANT REVERSE OSMOSIS: STAGE 1 COD
REJECTION 93
50. PILOT PLANT REVERSE OSMOSIS: STAGE 2 COD
REJECTION 94
51. PILOT PLANT REVERSE OSMOSIS: STAGE 3 COD
REJECTION 95
52. PILOT PLANT REVERSE OSMOSIS DATA: OXYGEN
DEMAND REDUCTION 97
IX
-------
FIGURES (CONT.)
Page
53. SIMPLIFIED FLOW DIAGRAM: ULTRAFILTRATION
SYSTEM 106
54. SIMPLIFIED FLOW DIAGRAM: REVERSE OSMOSIS
SYSTEM 110
-------
TABLES
Paqe
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
AVERAGE COMPOSITION OF WHEY PRODUCTS
ANALYTICAL PROCEDURES
RETENTION OF LOW MOLECULAR WEIGHT SOLUTES
IN ULTRAFILTRATION
MICROORGANISMS IN COTTAGE CHEESE WHEY PROTEIN
REVERSE OSMOSIS LIFE DATA: PERFORMANCE OF
DIFFERENT RO UNITS
ULTRAFILTRATION SECTION COMPONENTS
REVERSE OSMOSIS SECTION COMPONENTS
SAMPLING POINTS FOR ULTRAFILTRATION SECTION
SAMPLING POINTS FOR REVERSE OSMOSIS SECTION
ANALYSES USED FOR PILOT PLANT PROGRAM
PILOT PLANT ULTRAFILTRATION DATA
COD REJECTION DISTRIBUTION OF MODULES IN
STAGE 1
BACTERIAL DATA: TOTAL PLATE COUNTS AND
E. COLI COUNTS
PROJECTED CAPITAL COSTS FOR 300,000 LBS
WHEY/DAY PLANT
PROJECTED OPERATING COSTS FOR 300,000 LBS
WHEY/DAY PLANT
PUBLICATIONS AND PAPERS
VISITORS TO WHEY PILOT PLANT AT CROWLEY'S
6
19
23
37
53
63
67
72
72
73
78
96
100
111
113
116
MILK COMPANY 118
xi
-------
SECTION I
CONCLUSIONS
1. Membrane processes offer an economically attractive sol-
ution for waste treatment of cottage cheese whey. Pilot op-
eration has shown that a two-step process sucessfully pro-
duces protein concentrates through whey fractionation by ul-
trafiltration and lactose concentrates through reverse osmo-
sis .
2. The BOD of raw cottage cheese whey can be reduced from
approximately 35,000 mg/1 to less than 1,000 mg/1 by mem-
brane processing.
3. Protein concentrates containing up to 20% protein (80%
protein on dry solids bv.sis) can be generated in a single
step ultrafiltration.
4. Lactose concentrates containing 20% lactose (75% lac-
tose on a dry solids basis) can be generated by concentra-
tion through reverse osmosis of the ultrafiltration permeate.
5. Whole cottage cneese whey can be concentrated to about
25% solids by reverse osmosis without prior deprote•'.,-: i zation
(by ultrafiltration or otherwise).
6. Products produced from the two-step membrane process,
protein and lactose concentrates, have low total plate counts
and nil coliform counts. Products produced in the pilot
plant operation met standards for Grade A milk.
7- Concentration of whole whey by reverse osmosis may be more
difficult than concentration of deproteinized whey because
clean-up and sanitizing of the membrane equipment is complica-
ted by the presence of casein fines and protein.
8. Installation and operation of a 300,000 Ibs/day plant
are expected to yield a high return on investment before
taxes, co:. .esponding to a rapid plant pay-out. This conclu-
sion incorporates both capital and operating costs, as well
as information proprietary to Crowley's Milk Company regard-
ing the utilization and value of protein and lactose pro-
ducts .
9. Plant capital cost for a 300,000 Ibs whey/day installa-
tion is projected to be $697rOOO; and annual operating :. ^st,
$220 ,000.
-------
SECTION II
RECOMMENDATION
Crowley's Milk Company recommends the installation and oper-
ation of a 300,000 Ibs cottage cheese whey/day membrane pro-
cessing facility at the LeFargeville, N.Y. plant. This unit
is to be constructed in Phase II of the program.
-------
SECTION III
INTRODUCTION
The status of the pollution problem resulting from the dis-
posal of acid cheese whey is such that new methods must be
used to deal with it effectively. This section of the re-
port presents a summary of the size and nature of the waste
problem and provides a general description of the membrane
separation process under demonstration.
BACKGROUND AND NEED FOR PROJECT
The manufacture of cheese from either whole or skim milk
produces, in addition to the cheese itself, a greenish-
yellow fluid known as whey. Whole milk is used to produce
natural and processed cheeses such as cheddar, and the re-
sulting fluid by-product is called sweet whey, with a pH in
the range of 5 to 7. Skim milk is the starting material
for cottage cheese and gives a fluid by-product called acid
whey, with a pH in the range of 4 to 5. The lower pH is a
result of the acid developed during or employed for coagu-
lation.
Each pound of cheese produced results in five to ten pounds
of raw fluid whey. The high organic content of whey leads
to a severe disposal problem. Over 70% of the nutrients from
skim milk show up in acid whey, including proteins and lac-
tose. These materials, if properly recovered, could provide
useful products. Table 1 shows typical compositions of whey
and dried whey solids.
Production of Cottage Cheese Whey:
Some statistical information on the production of cottage
cheese (acid) whey is summarized here to provide a basis
for identifying the needs for acid whey treatment. In
categorizing industries in the United States, cottage cheese
is considered to be part of fluid milk rather than cheese
production. Thus statistics on acid whey are somewhat more
difficult to obtain than those for cheddar and other sweet
wheys. However, the general trends in the fluid milk indus-
try are the same as those for the natural cheese industry,
i.e., the industry is growing, but the number of plants is
decreasing. These trends are shown in Figure 1. (1) The
trend towards exceptionally large regional plants with dis-
tribution over wide areas is particularly notable.
-------
TABLE 1
AVERAGE COMPOSITION OF WHEY PRODUCTS
Nitrogenous
Product Water Matter Fat Lactose Acid Ash
Nitrogenous
Matter
0.7-0.9%
7-8%
12-14%
Fat
0.05-0.6%
1-2%
0.3-5.0%
Lactose
4.5-5.0%
28%
65-70%
Acid
0.2-0.6%
1-3%
2-8%
Raw Cheese Whey 93-94% 0.7-0.9% 0.05-0.6% 4.5-5.0% 0.2-0.6% 0.5-0.6%
Condensed Whey 50-60% 7-8% 1-2% 28% 1-3% 5-6%
Dried Whey 2-6% 12-14% 0.3-5.0% 65-70% 2-8% 8-12%
-------
By 1969, overall production of cottage cheese curd and
creamed cottage cheese had risen to an adjusted value of
966 million pounds. The acid whey output accompanying this
value was set at 4.83 billion pounds, based on 5 pounds whey
per pound of cottage cheese. A present output of at least
5 billion pounds of acid whey per year is considered to be
a conservative value, based on the current rate of growth
of cottage cheese production and on the fact that statistics
for other related acid cheese wheys (bakers, farmers, cream)
are not included.
The Cottage Cheese Whey Disposal Problem:
The organic nutrients of whey, which go unused, place a
costly burden on sewage systems and waterways. The bio-
logical oxygen demand (BOD) of whey ranges from 32,000 to
60,000 ppm. Most of this BOD is due to the lactose. Spe-
cific BOD values for cottage cheese wheys are between 30,000
and 45,000 mg/1, depending primarily on the specific cheese
making process used.
Every 1000 gallons per day of raw whey discharged into a
sewage treatment plant can impose a load equal to that from
1800 people.This is partially passed into streams in most
cases because BOD removal is not complete. Every 1000 gal-
lons of raw whey discharged into a stream requires for its
oxidation the dissolved oxygen in over 4,500,000 gallons of
unpolluted water.
Recent statistics give BOD's of about 0.2 pounds per pound
of cottage cheese curd. Combining this with the production
statistics given above, a total BOD removal of over 200 mil-
lion pounds is required for complete waste treatment. Con-
sidering that at the very best only half of the cottage cheese
whey produced is currently put to good purpose and that curd
wash water can contain 3% solids, the disposal problem is
severe.
PROPOSED MEMBRANE PROCESS
A two-step membrane separation process has been developed
for the treatment of cottage cheese whey. Although the
demonstration has been for acid whey treatment, the tech-
nology is applicable to sweet wheys. In this process, whey
is simultaneously fractionated and concentrated to give
protein and lactose by-products. The final effluent has
a low biological oxygen demand (BOD) and is expected to be
suitable for reuse within the cottage cheese plant. One
application, for example, is in curd washing. If the ef-
fluent is discharged, a final treatment for residual BOD
removal may be required, depending on local and state regulations,
-------
70
60
50
pRnpnrTinN (BILLION i RS.)
GO CD
en en
LA
LO
en
CD
UD
en
LPi
UD
en
NUM
8,000
5,000
2,000
ER OF PLANTS
OO CD
^r i_n
en en
un
en
CD
UD
en
un
UD
en
CD
r-x
en
LO l\
en en
CD
h«x
en
en en
30
PRODUCTION PER PLANT (MILLION LBS.)
20
10
OO CD
en en
LO
LO
en
CD
UD
en
to
en
CD
rv.
en
en en
FIGURE 1; TRENDS IN FLUID MILK PRODUCTION (1)
8
-------
The two-step whey treatment process is based on the applica-
tion of ultrafiltration (UF) and reverse osmosis (.RO) . Re-
verse osmosis has been extensively studied during the past
ten years under funding from the Office of Saline Water,
the bulk of this work focusing on desalination of brackish
and sea waters. Additional publicity has been given to its
application to whey concentration by the United States De-
partment of Agriculture. (_3,J,J>) Ultrafiltration is a varia-
tion of this membrane separation technique.
Figure 2 shows the basic concept involved. A semi-permeable
membrane separates water and a solution. In the absence of
a hydrostatic pressure differential, water will permeate the
membrane so as to dilute the solution. A counter pressure
can be applied to the solution side to reverse water trans-
port. The amount of pressure required to achieve a static
equilibrium is termed the osmotic pressure. Reverse osmosis
is simply the application of a pressure greater than the os-
motic pressure which drives water from the solution side of
the membrane to the water side, and permits concentration
of the solution.
The membrane plays the most critical role in the process.
By varying the properties of the membrane, one can control
the retention or passage of selected solutes. When the sol-
ute molecules are large, for example whey proteins, the os-
motic pressure is quite low and a membrane with relatively
large pores can be used at low operating pressures: e.g.,
10 to 100 psi. This is termed ultrafiltration. Referring
to Figure 3, the membrane may be called "loose"; that is,
lower molecular weight solutes will pass through the mem-
brane and will not be retained in the concentrate.
In reverse osmosis, solutions of small molecules with mod-
erate to high osmotic pressures are retained, and the re-
quired driving force is considerably higher, ranging from
several-hundred psi to over one-thousand psi. Higher opera-
ting pressures are required because of the substantial osmo-
tic pressure of salt and sugar solutions, and also because
of the greater resistance to water transport of RO membranes.
Again referring to Figure 3, a "tight" membrane is
employed.
Figure 4 shows a simplified flow sheet for the two-step
whey treatment process. Cottage cheese whey, with or with-
out filtration for fines removal, is introduced into a low
pressure UF unit (step 1). In this operation, whey is con-
centrated 10 to 30-fold by volume. Ultrafiltration membranes
are used which retain only the whey proteins. Thus, it is
-------
OSMOSIS
1 ATM
H20
SOL'N
1 ATM
REVERSE OSMOSIS
1 ATM
HLO
-T-SOL'N
PRESSURE
GREATER THAN
SOLUTION OSMOTIC
PRESSURE
FIGURE 2
OSMOSIS AND REVERSE OSMOSIS
-------
D o
SKIN
~ 0.3u THICK
POROUS
SUPPORT
SKIN
0.3y THICK
POROUS
SUPPORT
FIGURE 3
MECHANISM OF MEMBRANE SEPARATION
-------
possible to obtain a protein concentrate with a higher pro-
portion of proteins in the dissolved solids, since lactose,
non-protein nitrogen, lactic acid, and minerals pass through
the membrane. Operation is typically in the pressure range
of 10 to 100 psi, and at temperatures of 60 to 130°F. As
will be demonstrated in the data presented below, the pro-
tein content of raw whey can be increased from an initial
value of about 0.6% up to levels approaching 20% in this
step. Simultaneously, the proportion of protein in the
whey solids can be increased from approximately 10% to 80%
in a one-step concentration.
The permeate (ultrafiltrate) from the UF unit is introduced
into a second membrane step. In an RO operation, this
stream is concentrated from approximately 6% solids to 20-
25% solids. Typical operation would be in the pressure
range, 500 to 1500 psi, and at a temperature of 60 to 100°F.
The membrane in the RO section is chosen so as to retain
as great a proportion of the organic solutes as possible,
resulting in the permeate having a low BOD. The final ef-
fluent from the RO section can either be reused within the
dairy or cheese plant or discharged, with or without treat-
ment for residual BOD removal. Where pollution control re-
gulations are not overly stringent, a moderate salt rejec-
tion membrane can be used in the RO section to permit par-
tial desalting and lactic acid removal from the lactose
concentrate. In general, from the point of view of pollution
control, this option will probably not be exercised.
The protein concentrate can either be used directly by in-
corporation into food products, or it can be dried. Drying
may be preceded by concentration by vacuum evaporation. The
lactose concentrate can be further concentrated by evapora-
tion and the lactose can be recovered in a simple crystal-
lization operation.
Another process alternative, which has not been explored in
depth in this program, is the direct concentration of whole
whey by RO. This approach has been examined by others
(_2,_3,£) , and appears to be attractive for cheese producers
with small volumes of whey. Concentration on site would
precede transportation to, and processing at, a central
facility- However, a satisfactory cleaning program needs
to be defined prior to wide use.
In the immediately following sections, data for the operation
of the UF and RO sections will be presented. TRfe data pre-
sented first were obtained with prototype systems operated
at Abcor, Inc., Cambridge, Massachusetts. The latter data
are from operation with a 10,000-to-20,000-lbs whey/day
pilot plant operated at Crowley's Milk Co., Binghamton, N.Y.
12
-------
PROTEIN
WHEY
CONCENTRATE
STEP 1^-
UF
(jO
10-100 PSIG
LACTOSE
Hz 0, LACTOSE
NON-PROTEIN N.
LACTIC ACID
SALTS
CONCENTRATE
STEP 2
RO
>200 PSIG
LOW BOD
WATER
FIGURE 4 -MEMBRANE PROCESS FOR WHEY
TREATMENT —FLOW SCHEMATIC
-------
SECTION IV
PROTOTYPE EXPERIMENTS
Extensive prototype experiments were performed prior to
the design, installation and operation of the pilot plant.
These were conducted at Abcor, Inc. from November, 1969
through June, 1970, under subcontract to Crowley's Milk
Co. In the course of these experiments, a variety of
membranes and membrane equipment were examined for both
UF and RO. Significant experimental results are presented
and discussed in this section.
UL TRAPIL TRATION
Experimental Procedure
Apparatus and Procedure. Two .systems were used to gener-
ate the data which follow. The first, containing 13.2
ft2 of Abcor HFA-180 tubular membranes, was used to
batchwise concentrate cottage cheese whey. A flow schema-
tic is shown in Figure 5. Cheese whey was charged to the
feed tank and heated to the desired operating temperature
by passing hot water (150°F) through a coil installed in
the tank. Temperature was maintained with a temperature
switch, installed in the tank, which actuated a solenoid
valve controlling hot water flow. With this system it
was possible to control the temperature of the whey to
within 4°F.
Whey was circulated through the tubular membranes with a
2 HP centrifugal pump. The system circulation rate was
measured with a rotameter and controlled with two globe
valves. These valves were also effective for regulating
the system operating pressure. Membrane system inlet and
outlet pressures were measured with diaphragm-protected
gauges. Fractional water removal per pass through the
membranes was very small, typically less than 0.001, and
the "conversion per pass" was negligible. Corresponding-
ly, the system contents were well-mixed. Permeate was
withdrawn continuously, while the concentrated feed was
recovered only at the end of the test, by draining the
tank, membranes and piping. This is termed a "batch con-
centration. "
A similar system was used for "differential" studies.
In this prototype system permeate was returned continu-
ously to the feed tank in order to maintain a time-in-
variant feed composition and concentration.
15
-------
TPMPFDATI IRF
1 Cl Ir CrvM 1 UK U
[" SWITCH
£>
fsj^l MOT WATFR
P^ (150'F)
_ _^
FEElD
TANK
JcOIL
,—^-1 r
— T D Y l
PUMP
^r* 1
^k -
ROTAMETER Q
PRESSURE PRESSURE V
9 9 T
^^1
•^Xl MEMBRANES
PERMEATE .
(TO STORAGE)
FIGURE 5
TEST SYSTEM FLOW SCHEMATIC
-------
Batch concentration experiments were employed to deter-
mine the dependence of membrane flux and membrane reten-
tion on feed concentration. Tests of this nature were
generally carried out at a fixed temperature, feed cir-
culation rate and pressure. Differential experiments
were performed to examine the effects of feed circula-
tion rate, pressure, temperature and time on membrane
performance.
Feed Whey. Cottage cheese whey was obtained from two
sources. For two experiments whey was shipped from Crow-
ley 's Milk Co., Arkport, New York. Unfortunately the de-
livery time of the whey was sufficiently long such that
spoilage of the whey was a problem. For this reason
whey for the other experiments was purchase from the H.P.
Hood & Sons Co. The whey was shipped from Newport,
Maine,immediately after collection and was processed
within 24 hours of shipment.
Typical properties of the whey are listed below.
pH 4.5
total acidity (expressed
as lactic acid) 0.5-0.65%
lactose 4.5-5%
protein (Kjeldahl) 0.7-0.8%
fat (Roese-Gottlieb) 0.02-.045%
total solids (Mojonnier) 6.2-6.6%
total plate count
(when shipped) 500-5,000 org/ml
The cottage cheese whey was processed in the initial ex-
periments without filtration. For later experiments
the whey was filtered with a 5y filter prior to intro-
duction into the membrane system. Filtration for fines
removal appeared to have no effect on membrane perfor-
mance, but may be required when a protein concentrate
free of casein fines is desired.
Analytical Procedures. Several analyses were used in the
course of the prototype tests. These are listed in
Table 2, where references for the procedures and their
sensitivities are indicated. Analyses performed at
17
-------
Abcor, Inc., included -COD, Lowry protein, lactose, pH,
total acidity and total solids. Analyses for Kjeldahl
protein (nitrogen x 6.3) and BOD were carried out at
Metcalf & Eddy, Inc., in Boston, Mass. The gel permea-
tion chromatography assays, and the viable microorganism
counts were performed at the Department of Nutrition and
Food Science at M.I.T-, by Professors Stephen R. Tannen-
baum and Anthony J. Sinskey, respectively.
For the UF experiments , key factors included the mem-
brane protein retention and the obtainable level of
protein concentration in the concentrate. Protein con-
tents presented in this section are Lowry "protein",
which was shown to be equivalent to Kjeldahl "protein",
based on analysis of identical samples at Abcor and
Metcalf and Eddy.
The lactose analysis was also a colorimetric analysis.
Standard curves were generated for lactose samples with
and without the addition of 6-lactoglobulin. The curve
without $-lactoglobulin was used to analyze for per-
meate samples from UF experiments (as well as for the RO
tests) . The curves with added 8-lactoglobulin (at dif-
ferent concentration levels) were used for the analysis
of whey and whey concentrate samples .
Abcor 's COD analysis was correlated with BODs analyses.
A series of samples were frozen and sent to Metcalf and
Eddy for analysis for 6005. The same samples were ana-
lyzed at Abcor. The correlation between BOD5 and COD
was. found to be, approximately , BOD=0.78 COD.
RESULTS
Membrane Retention Properties . The retention properties
of the ultraf iltration membranes have been examined in
detail. Retention is defined by Equation (1) :
R(%) = CF " CP x 100 (1)
CF
where CF is the feed protein concentration, an$ Cp is
the permeate protein concentration at the corresponding
feed concentration. This definition of retention is to
be contrasted with yield. Yield is defined as the frac-
tion of protein in the feed whey contained in the whey
concentrate. This is given by Equation (2) :
18
-------
TABLE 2: ANALYTICAL PROCEDURES
ANALYSIS FOR;
PROCEDURE:
SENSITIVITY;
COD
vo
BOD i
Protein (Lowry)
Protein (Kjeldahl)
Protein (Gel Permea-
tion Chromatography)
Lactose
PH
Total Acidity
Total Solids
Viable Bacteria
Std. Methods for the Exam.
ot Water, Sewage, and indT
Was tes, Amer. Pub. Health
Assoc., 12th ed., 1965
(Metcalf & Eddy)
J. Biol. Chem., 193, 265(1951)
(Metcalf & Eddy)
Analyzed on BioGel P10 Column
Biochem, J., 45, 445(1949)
Standard pH Meter
Titration, Standard Methods
Gravimetric at 105°C
Standard Plate Count at 37°C
0.05%
0.1%
0.05%
0.05%
0.1%
0 .05 pH unit
0.05%
0.05%
50 org/ml
-------
Y(%) = CC ' VC x 100 (2)
CF • VF
where Cc and Vc are the protein concentration in and vol-
ume of the concentrate, and CF and VF are the correspond-
ing values for the feed whey.
Although yield is the important quantity in evaluating
process performance, retention is a more meaningful way
to examine membrane properties. For this reason, the
data are presented in the form of retention. This is
also a more convenient form, since it eliminates consid-
erations pertaining to the degree of volumetric concen-
tration of the feed material.
Protein Retention. In batch concentrations the Lowry
protein retention was determined as a function of the
protein content of the feed. Data are given in Figure 6
for five tests. Protein retention increased rapidly
with increased feed concentration. This is due to the
fact that the Lowry protein analysis included both true
protein and non-protein nitrogen. As the feed became
more and more concentrated the proportion of true pro-
tein, and consequently the Lowry protein retention, in-
creased.
For all runs performed over a six-month period, protein
retention remained unchanged.
The "true" protein retention of the membranes was deter-
mined by gel permeation chromatography. Feed and per-
meate samples were fractionated on a Biogel P-10 col-
umn (column diameter 1.5 cm, and column length, 90 cm).
Ultraviolet absorbance of the column effluent was moni-
tored as a function of effluent volume. Data are shown
in Figure 7. The first peaks eluted from the column
corresponded to proteins, which were excluded from the
porous packing. Protein peaks were identified by cali-
brating the column with pure samples of soluble 6-lacto-
globulin and a-lactalbumin. The later peaks were other
UV absorbing materials. The traces in Figure 7 are for
feed whey and its corresponding permeate. It^is imme-
diately apparent that peaks corresponding to me higher
molecular weight globulins, 3-lactoglobulin, and a-lact-
albumin, were absent in the permeate sample, except for
a very small lactalbumin peak. On an area basis, about
98.5% of the protein was retained. Similar traces have
20
-------
(O
1 0 0
O
H
LU
LU
LU
I-
O
a.
90
9'
a
8 0
7 0
1 1 1 1 I
-O —O O-
-A ' tr
-o—o-
o EXP, # 15
o EXP, # 18
9 EXP, # 13 & 21
a EXP, # 26
16
T±
21T
* FEED PROTEIN
FIGURE 6
ULTRAFILTRATION DATA: PROTEIN RETENTION
-------
0*6
to
IO
UJ
o
o:
o
m
e-LACTOGLOBULIN
BIOGEL P-10 COLUMN
BLUENT - 0.05M NH4AC. PH 4.5
INITIAL FEED
INITIAL PERMEATE
\ LACTALBUMIN
AMI NO ACIDS AND OTHER LOW MW MATERIALS
200
ML EFFLUENT (CORRECTED FOR VOID VOLUME)
FIGURE 7
ULTRAFILTRATION DATA: GEL PERMEATION CHROMATOGRAPHY RESULTS
-------
been generated for whey protein concentrates and corres-
ponding permeates. These show similar behavior, with
protein retentions of 98 to 99%.
Both the Lowry protein retention data and the gel permea-
tion chromatography data indicated protein retention with
HFA-180 membranes was satisfactorily high. On this basis,
no further membranes were evaluated for the UF step.
Retention of Other Whey Solutes. Rejection of other sol-
utes in the ultrafiltration experiments was very low.
The data have been reduced to the form presented in Table
3.
TABLE 3
RETENTION OF LOW MOLECULAR WEIGHT SOLUTES
IN ULTRAFILTRATION
Component % Retention
non-protein nitrogen ^0
lactose 0
salts 0
lactic acid 10-20%
Retention of non-protein nitrogen, lactose and salts was
approximately 0%, and for lactic acid, very low. Thus,
the membranes used for the prototype experiments very ef-
fectively retained protein while passing lower molecu-
lar weight solutes. This resulted in the production of
very high protein content products through whey concentra-
tion to high volumetric reduction ratios. Date are shown
in a reduced form in Figure 8. Composition of the protein
concentrate on a dry solids basis is shown as a function
of the degree of water removal. At water removal levels
approaching 98%, protein concentrate streams can be gener-
ated with 80% protein content on a dry basis. Concentrates
of this composition were generated in the course of pro-
totype tests.
Membrane Flux. A key economic parameter in determining
capital and operating costs for membrane processes is the
rate of solution transport through the membrane (flux).
For the ultrafiltration of proteins and other macromole-
cular solutions flux has been correlated with certain sys-
tem parameters, expressed in Equation 3.
23
-------
CO
Q
O
V)
ce
Q
100
9 0
8 0
7 0
6 0
5 0
3 0
2 0
10
% % PROTEIN TO
WATER TOTAL LACTOSE
REMOVED SOLIDS RAJouDSRY
0
50
75
95
98
6.5-7.3
7.5-8.3
9.5-10
17-18
24-25
0.13-0.15
0.20-0.25
0.40-0.50
1.9-2.2
3.9-4.5
ASH
50 60 70 80 90 100
% WATER REMOVED BY ULTRAFILTRATION
FIGURE &
COMPOSITION OF WHEY PROTEIN CONCENTRATE
24
-------
-
J = k In ? (3)
This relationship — relates flux (J) to the mass trans-
fer coefficient in the membrane-flow system (k) , and
certain concentration terms . C* is a concentration de-
termined experimentally, Cp and Cp are feed and permeate
concentrations. The quantity (Cp-Cp) is the "retained"
protein concentration.
Flux is expected to increase in proportion to the mass
transfer coefficient in the membrane system and to de-
crease in proportion to the log of the retained protein
concentration .
In a turbulent-flow membrane system, the type of system
under consideration here, the mass transfer coefficient
can be correlated with additional system parameters. A
generally accepted correlation is given in Equation 4 . ($)
k = 0.0091 Re°-91Sc-°-65 (4)
The mass transfer coefficient is related to the operating
Reynolds number in the system (Re) , the solution\Schmidt
number (Sc) , the solution kinematic viscosity (v) , and
the tubular membrane diameter (d) . The Reynolds number
in turn, is determined by the circulation rate in the sys-
tem (G) , the tube diameter, and the solution kinematic
viscosity by Equation 5.
Re = 4G (5)
irdv
The Schmidt number is. the ratio of the kinematic viscos-
ity to the protein diffusivity, (v/D) .
Equations (3) and (4) will be used to correlate flux
data.
Effect of Time on Flux (short term tests) . The effect of
time on membrane flux was examined in experiments lasting
a few days. For well-cleaned membrane systems, a small
decline in flux over the first 4 to 8 hours was generally
observed. This is indicated by the data in Figure 9, in
which flux in gallons/day-f t2 (gfd) is shown as a func-
tion of operating time. Data are presented from two
25
-------
to
2 5
2 0
O EXP, # 1
A EXP, # 11
Q
a.
X
_l
u.
1 5
10
-A-
10
2 0
3 0
't 0
5 0
6 0
7 0
OPERATING TIME/ HOURS
FIGURE 9
ULTRAFILTRATION DATA: DEPENDENCE OF FLUX ON TIME
(0.62% FEED PROTEIN; eo0F; is PSI; IB GPM CIRCULATION RATE)
8 0
-------
experiments. Operation was at 15 psi, 60°F, and 18 gpm
circulation rate. In this differential experiment, with
all permeate being returned to the feed reservoir, the
Lowry protein content of the whey was maintained at a
level of 0.6%.
It was not determined if the decrease in flux was due to
deterioration of the whey in the membrane system, possi-
bly related to the growth of microorganisms, or the form-
ation of a protein precipitate on the membrane surface.
However, this short-term flux decay with time was observed
in both prototype and pilot-plant experiments.
Regardless of the cause of this decline the cleaning pro-
cedure developed in the course of prototype experiments
was satisfactory in that initial whey fluxes could always
be recovered after cleanup.
Effect of Whey Circulation Rate. According to Equations
(3) and (4), UF flux is expected to increase with the 0.9
power of circulation rate. This is confirmed by the data
shown in Figure 10. Flux in gfd is shown as a function
of circulation rate (gpm), for both feed whey (0.6% pro-
tein) and about 6-fold concentrate (3.3% protein). The
lines drawn through the data for both experiments have
slopes of approximately 0.9, in excellent agreement with
theory.
Effect of Operating Pressure. According to Equation (3),
flux should be independent of operating pressure, since it
is limited by fluid-phase mass tranfer in the membrane
system and not membrane resistance. This has been con-
firmed experimentally and typical data are shown in Figure
11. Over the pressure range of 20 to 40 psi, flux was in-
dependent of the operating pressure.
Effect of Temperature. The major effect of temperature
will be on the mass transfer coefficient k. An increase
in temperature will decrease the whey viscosity and in-
crease the protein diffusivity in solution. These two
effects will lead to an increase in the mass transfer
coefficient. Data illustrating this are shown in Figure
12. Flux has been plotted against the reciprocal of
temperature in degrees Kelvin. The data cover the temp-
erature range of 60 to 110°F. This Arrhenius plot is
linear and lines drawn through the data correspond to ac-
tivation energies of 5.1 and 5.5 kcal/mole.
For flux considerations, an optimum operating temperature
27
-------
2 0
15
Q 10
.^ SLQPE s 0.96
L 1.1,1
it 6810 20 tO
CIRCULATION RATE/ GPM
FIGURE 10
ULTRAFILTRATION DATA:
DEPENDENCE OF FLUX ON CIRCULATION RATE
O EXP, # 7; 0.6% FEED PROTEIN;
60°F; 20 PSI
A EXP, # 11; 3.3% FEED PROTEIN;
79°F; 12 PSI
28
-------
to
vo
2 8
2 0
Q
U_
16
12
A
A
A
A
A 0.85% FEED PROTEIN; 95°F; 21 GPM CIRCULATION RATE
O 2.9% FEED PROTEIN/' 78°F; 13.2 GPM CIRCULATION RATE
o
i o
3 0
PRESSURE/ PSI
FIGURE 11
ULTRAFILTRATION DATA: DEPENDENCE OF FLUX ON PRESSURE
5 0
-------
U)
Q
1 5
1 0
CO
LL.
3.12
3.20
ACTIVATION ENERGY
5.5 KCAL/MOLE
ACTIVATION ENERGY
5.1KCAL/MOLE
I
I
3.28
3.36
3.41*
1/T X 103/ ("K)"1
FIGURE 12
ULTRAFILTRATION DATA: DEPENDENCE OF FLUX ON TEMPERATURE
D EXP, # 15; 3.2% FEED PROTEIN; 38 PSI; 13.2 GPM CIRCULATION RATE
o EXP, # 11; 4% FEED PROTEIN; 12 PSI; 14.3 GPM CIRCULATION RATE
-------
appears to be about 125 F. For cottage cheese whey this
minimizes plant heat exchange requirements since whey is
received hot from the cheese vats. Preliminary results,
not described in this report, show little if any heat de-
naturation of the whey proteins at this temperature. Fur-
thermore, as will be discussed, the rate of bacterial
growth at 125°F appears to be low.
Effect of Concentration. According to Equation (3), flux
is expected to decrease logarithmically with increasing
feed concentration.
Data shown in Figure 13 support this hypothesis. Flux is
plotted versus log of the retained Lowry protein concen-
tration for operation between 61°F and 120°F. Although
there is some scatter, the data can be fit by straight
lines on this semi-log plot. As is expected, increased
temperature resulted in increased flux levels. From the
intercept of the lines, the quantity (C*-C_) in Equation
(3) is estimated to be 30%.
Life Tests. Life tests have been in progress with 14 tu-
bular membranes since September, 1969. Data for flux and
protein retention over the ensuing 14 months are shown in
Figure 14.
Standard procedure has been to operate with a single batch
of whey for one week, followed by clean-up (see page
for procedure) and operation with a new batch. Fluxes re-
corded were measured within 10 to 24 hours after introduc-
tion of fresh whey. In the course of operation, over a
5 to 7 day period, substantial microbiological growth
was observed in the membrane system. This was manifested
in very high counts of yeasts, molds, and thermophiles.
In spite of this very extensive and regular microbiologi-
cal contamination, no detectable deterioration of the mem-
branes occurred.
Flux for raw whey, with or without filtration, averaged
about 20.5 gfd, while membrane protein retention varied
between 80 and 90%. Scatter in protein retention is
probably due to variation in the cheese whey. It is ap-
parent that no detectable deterioration occurred in either
membrane flux or protein retention, demonstrating membrane
life in excess of one year. These results illustrate the
dependability of the low-pressure ultrafiltration system,
and have been confirmed in the pilot plant operation.
During the life tests, the membranes were exposed to whey
31
-------
10
to
2 0
i e
1 6
11*
Q 12
u_
o
% i o
U_ 8
77 F; 13.2 GPM
0 EXP, # 11; 61 F; 10 GPM
EXP, # 18; 102°F; 13 GPM
EXP, # 21; 77°F; 13.2 GPM
EXP, # 26; 120°F; 13.5 GPM
5 GPM
F; 13.0 GPM
61 F; 10 GPM
0.1 0.2
0.51 2 5
% RETAINED PROTEIN/ (cp - Cp)
i o
20 30
FIGURE 13
ULTRAFILTRATION DATA: DEPENDENCE OF FLUX ON PROTEIN CONCENTRATION
-------
LO
U>
--O
III •—•
J_J_
O2
QCUJ
Q-J-
LU
1 0 0
80
.0-5 0—8
00
-
SEPT OCT NOV DEC JAN FEE MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC
1969 1970
3 0.
20
1 0
0
°° ° o o =8
«o a „ (,„„ „ . . 0°
"ft O O O Oft ••> O yo -
0 u o D 0
8 o o
P o 0 0
. 1 1 1 1 1 1 1 • . 1 1 . , 1 1 1 1 . 1 J . I.I II 1 1 1 1 1 . 1 1 ... 1 1 1 1 . 1 . . 1 1 1 1 1 . . 1 1 , 1 . 1 1 . . . 1 I 1 ... 1 . . 1 1 . 1 . 1 . . 1 1 1 . . .
SEPT OCT NOV DEC JAN FEE MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC
1969 1970
FIGURE 14
ULTRAFILTRATION LIFE DATA; FLUX AND PROTEIN RETENTION
(COTTAGE CHEESE WHEY WITH AND WITHOUT FILTRATION AS FEED;
0.8% PROTEIN; 15 GPM CIRCULATION RATE; H3°F; 25 psi)
-------
under operating conditions slightly more 'than 50% of the
time. The remaining time was taken up by water tests, sys-
tem cleaning and sanitizing, shutdown due to non-availabil-
ity of whey, or for vacations and holidays. During the
life tests, there was no case of membrane failure.
Cleaning and Sanitizing.
Procedures. In the course of the long-term experiments,
it became apparent that membrane fouling occurred. This
was evidenced by formation of "scale" on the membrane sur-
faces, resulting in reduced water and whey fluxes. In ad-
dition, microbiological contamination of the system became
severe in the absence of cleaning and sanitizing. For
these reasons, a variety of dairy cleaners and sanitizers
were examined. Important considerations were the follow-
ing:
-ability of cleaners to remove scale from the mem-
branes, resulting in recovered water flux;
-absence of chemical attack of the membranes by
cleaning solutions;
-effective sanitization of the membranes by sani-
tizers;
-absence of chemical attack of the membranes by
sanitizers.
A variety of acid, neutral, and alkaline detergents, both
ionic and non-ionic, were examined. In addition, several
sanitizing agents, including formaldehyde, iodofores, hy-
drogen peroxide, and chlorine, were tested. Although
many of the cleaners and sanitizers were effective, the
following cleaning and sanitizing sequence was adopted
early in the life tests program because of its simplicity,
effectiveness, and acceptability to the dairy industry.
Cleaning. Following operation with whey, the membrane
system was flushed with tap water. Next, a cleaning sol-
ution consisting of 1 oz Alcozyme per 2 gals of water was
circulated through the membrane system at 40-50°C. Alco-
zyme is an alkaline enzyme detergent manufactured by the
Alconox Co., N.Y. The pH of the cleaning solution was
adjusted to 9.1-9.3, if originally above this level. Ad-
justment was made with HCl. The cleaning solution was
circulated through the membrane system at 15-25 gpm, for
30 min to 1 hr. During this period, permeate was returned
34
-------
to the cleaning solution tank. Following cleaning, the
membrane system was again flushed with tap water, and
stored filled with water until the next experiment. If
storage was to be longer than one week, the system was
sanitized prior to filling with water.
Sanitizing. Prior to operation with whey the membrane
system was sanitized. A solution of Antibac B (Wyandotte
Chemical Co.) at a concentration of 1 oz/12 gal water was
circulated through the membranes. This sanitizing solu-
tion provides 100 ppm free chlorine at a pH of about 5.
The sanitizing solution was circulated for 15-20 minutes
at 80-100°F and 10-20 gpm. As with the cleaning solution,
the permeate was returned to the santizing solution tank.
Following this operation, the sanitizing solution was dis-
placed from the membrane system with whey. Ultrafiltra-
tion experiments commenced as soon as displacement was
complete.
Microbiological Data. The effectiveness of the above
cleaning and sanitizing procedure is indicated by the fol-
lowing data.
In Figure 15 total plate counts of one batch of whey are
presented as a function of holding time in the membrane
system. The actual counts correspond to counts in sam-
ples taken as concentration proceeded during a batch ex-
periment. Starting at a level of slightly more than
1,000 org/ml, the total plate count increased to about
80,000 org/ml. Ultrafiltration membranes selectively re-
tain microorganisms (microorganisms have orders-of-magni-
tude higher molecular weights than whey proteins), and
these counts should be corrected for the volumetric con-
centration ratio to have a measure of actual growth rates.
This has been done as indicated. The actual growth rate
was indeed low.
In Table 4 are shown microbiological data for cottage
cheese whey protein produced in the prototype batch con-
centration system. A fluid protein concentrate was pro-
duced containing about 15% protein (70% protein on dry
solids basis). Total plate count of the concentrate was
74,000 org/ml. In freeze-drying the protein, this
dropped to 37,000 org/gm; and in spray drying the pro-
tein, this dropped to 600 org/gm. Apparently thermal
conditions in the spray drier were effective for killing
bacteria. The complete absence of yeasts, molds, and
coliforms is to be noted.
35
-------
1 0
UJ
en
CD
t£.
O
o
o
'LU
O
I-
10
id".- ACTUAL COUNTS
CORRECTED FOR VOLUMETRIC
CONCENTRATION
o
id 20
HOLDING TIME/ HOURS
3 0
FIGURE 15
ULTRAFILTRATION DATA: BACTERIAL GROWTH DURING WHEY CONCENTRATION
-------
Although the prototype system was of a non-sanitary design,
products were generated with microbiological counts accept-
able for food-grade products. This result has subsequently
been reproduced in the pilot plant operation.
TABLE 4
MICROORGANISMS IN COTTAGE CHEESE WHEY PROTEIN
(77% Protein in Dry Solids)
Fluid Freeze Dried Spray Dried
Concentrate Protein Protein
Total Plate Count
(TSY Agar) 74,000 org/ml 37,000 org/ml 600 org/ml
Yeasts and Molds
(Sabourauds Dex-
crose Agar) 0 0
Coliforms
(McConkey's Agar) 0 0
REVERSE OSMOSIS
Experimental Procedure and Results
Apparatus and Procedure. Two prototype systems were used
for RO experiments,similar in layout to that of the UF
systems (See Figure 5). Major differences include the use
of high-pressure pumps, valves and piping since operation
was at substantially higher pressures. Positive-displace-
ment piston pumps were used, with stainless steel accumula-
tors added to damp out flow pulsations. Although major
pulsations were not present "micropulsations" were, are
could have contributed in some degree to module failures
discussed below.
Experiments were of two types, differential and batch con-
centration. In differential tests the effects of pressure
and recirculation rate on flux and membrane rejection were
examined. In batch concentration experiments the effect
of feed concentration on membrane flux and rejection was
determined. A variety of membrane modules were examined,
including systems manufactured by Havens International,
American Standard, Universal Water Corporation, and Aero-
jet General. The number and types of modules are listed
in Table 5-(page 53).
37
-------
Analytical Procedures. The analytical procedures which
were used have been discussed in the section entitled
"Prototype Experiments: Ultrafiltration." These included
analyses of feed and permeate samples for COD, total sol-
ids, lactic acid and lactose. All of the experiments,
unless otherwise noted, were performed with UF permeate
as feed.
Membrane Rejection Properties. Rejection characteristics
have been determined for the modules listed in Table 5.
These data are presented in this section.
Solids rejection and solids level in the permeate are shown
as functions of feed concentration in Figure 16.
Data are given for several types of modules including Havens
Types 300 and 310, American Standard TM2-4 with AS-189
membranes, American Standard TM5-14 modules with AS-194
and AS-197 membranes, Universal Water Corp. modules, and
eight Aerojet tubular membranes. Total solids rejection
was observed to be best for the Aerojet tubes and the
AS-194 membranes, followed by the Universal Water Corp.
modules. The AS-197 membranes did not exhibit as high a
rejection as would be expected since these membranes
should have higher retention properties than the AS-194
membranes. Havens Type 300 and 310 modules showed the
lowest rejection.
Similar data for COD rejection are given in Figure 17.
COD rejection paralleled that for total solids. The
most retentive membranes, the American Standard AS-194
membranes and Aerojet tubes, had COD rejections exceeding
99%, and this is an obtainable level for plant operation.
Data for lactose rejection are given in Figure 18. Al-
though there was more scatter for these data, rejection
of lactose by the membrane modules was of the same order
as total solids and COD rejections.
The rejection of lactic acid by RO membranes was only mod-
erate. This is demonstrated by the data of Figure 19.
Rejection is shown as a function of lactic acid concentra-
tion in the feed. Typical rejection levels ranged be-
tween 80 and 90% with UF permeate as feed. Wjaen raw cot-
tage cheese whey was used as feed, rejection levels were
somewhat higher. This cannot be easily explained, but
may be due to the formation of a high-rejection dynamic
membrane by whey proteins.
38
-------
CO
Q
O
V)
dft
UJ
I-
<
LLJ
UJ
O.
1 _
2 0
FEED % SOLIDS
2 5
3 0
100
cfP
H
o
UJ 90
111
CC
CO
Q
80
CO
1
1
10 15 20
FEED % SOLIDS
2 5
3 0
FIGURE 16
REVERSE OSMOSIS DATA: PERFORMANCE OF DIFFERENT MODULES
(COTTAGE CHEESE WHEY - UF PERMEATE AS FEED)
v HAVENS 310 MODULE/ 575 PSIG
• UNIVERSAL WATER CORP, MODULE #349, 600 PSIG
o AMERICAN STANDARD AS-189, 710PSIG
• AMERICAN STANDARD AS-197, 570 PS.IG
A HAVENS 300 MODULE/' 740 PSIG
o UNIVERSAL WATER CORP- MODULE #395, 780 PSIG
O AEROJET TUBES/ 590 PSIG
a AMERICAN STANDARD AS-194, 580 PSIG
39
-------
1 0
*>
O
m
l
o
rH
Q_
a.
Q
o
o
<
UJ
2:
cc
UJ
HAVENS 300 & 310, AND
AMER, STD, AS-189-j
UNIV, WATER
& AEROJET TUBES;
,
AMER, STD, AS-194^
& lAERQJET TUBES i I
200
0 50 100 ISO
FEED COD/ PPM X 10"3
FIGURE 17
REVERSE OSMOSIS DATA: COD REJECTION
(SEE FIGURE 16 FOR LEGEND; EXCEPT* 590 PSI/ AEROJET TUBES)
-------
I-
<_)
LU
->
LU
LU
CO
O
I-
o
I 0 0
9 6
9 2
AMER, STD AS-194 &
AEROJET TUBES
^ u-
A * A *
* A jf --g- «
~ ^J ^ ^ O
B • /
• •HAVENS 300 & 310
* AMER, STD, AS-189
i i i i i i i
+ •
a \ •
<7 \ r? A
0 \ v "
\ o
• UNIV, WATER
& AEROJET
TUBES
i i i
LU
LU
D_
LU
CO
O
o
df>
10 12 14 16
2 0
% LACTOSE IN FEED
FIGURE 18
REVERSE OSMOSIS DATA: LACTOSE REJECTION
(SEE FIGURE 16 FOR LEGEND; EXCEPT + 590 PSI, AEROJET TUBES)
-------
Q
I—«
O
Q
i—i in
01-
«
LU
os:
t-> C£.
HLU
O Q_
<
%
to
100
90 -
80 -
7 0
0 . 2 _
0 . 1 _
05 1.0 1.5
% LACTIC ACID IN FEED
2.0
UNIVERSAL WATER CORP. MODULES
ORAW WHEY AS FEED/ 680 PSI/ 26'
• UF PERMEATE AS FEED/ 590 PSI/
AMERICAN STANDARD MODULES
C/ 2
25°C,
• RAW WHEY AS FEED/ 610 PSI,
OUF PERMEATE AS FEED/ 600 PSI/
£UF PERMEATE AS FEED/
• UF PERMEATE AS FEED/
GPM
1.8
GPM
26°C/ 2
25°C/
570 PSI/ 25 C/
570 PSI/ 25°C/
GPM/ AS-189 MEMBRANES
2 GPM/ AS-189 MEMBRANES
1.8 GPM/ AS-194 MEMBRANES
1.8 GPM/ AS-197 MEMBRANES
HAVENS MODULES
v UF
PERMEATE AS FEED/
WHEY AS FEED/ 650
570 PSI/ 25°C/
PSI/ 26°C/
1.8 GPM/ 310 TYPE MEMBRANES
2 GPM/ 300 TYPE MEMBRANES
FIGURE 19: REVERSE OSMOSIS DATA; LACTIC ACID REJECTION
-------
By comparison of Figures 17 and 18 it is seen that COD re-
jections are lower than lactose rejections. This is attri-
buted to the partial passage of lactic acid into the per-
meate. For this reason treatment of high-acid wheys will
result in a higher effluent BOD. It is desirable, there-
fore, to treat cottage cheese whey as soon as possible af-
ter it is removed from the cheese vats.
Effect of Pressure on Membrane Rejection. According to RO
theory, an increase in operating pressure will increase
membrane rejection efficiency. This is demonstrated by
the data in Figures 20, 21, and 22. Rejection data are
shown for the American Standard TM2-4 module with AS-189
membranes. Rejections for total solids, COD and lactose
are observed to increase substantially with increases in
operating pressure. These data indicate that operation
of the RO section should be at as high a pressure as pos-
sible in order to minimize the final effluent BOD. The
major limitation will be mechanical, with the maximum
operating pressure determined by two factors: membrane
compaction and mechanical strength of the membrane sup-
ports .
Membrane Flux; General. Reverse osmosis systems are gen-
erally designed so that concentration polarization is not
an important consideration. This is achieved by determin-
ing the circulation rate above which flux and rejection
are independent of circulation rate, and operating above
this level. Under this condition, membrane flux is de-
termined, by two parameters: the inherent membrane resis-
tance to water transport and the osmotic pressure of the
feed. Solution flux is given by ,
J = A(AP - ATT) (6)
where J is the flux; A is the membrane constant; and AP
and ATT are the differences in applied pressure and osmo-
tic pressure across the membrane. The flux of dissolved
materials (solutes) is described by the equation
J ., = BAG (7)
sol
where JSol is the flux of a dissolved component, B is the
permeation coefficient through ;the membrane, and AC is the
concentration gradient across the membranes. These two
equations predict increased rejection with increased oper-
ating pressure. This fact is a consequence of the nature
of the driving forces in RO. Thus, the experimental re-
sults presented above, showing rejection increasing with
pressure, are in accord with theory.
43
-------
SOLIDS REJECT-
CD U> O
O O O
800 PSI - o & £-A A-AC
A r| n— -
° D 650 PSI
0 oo ____
500 PSI
i I I
D 5 10 IS
FEED % SOLIDS
3 A A A —
_ — Q'
~i — * • •*
0_ 600 PSI
1
20 25
2 5
FEED % SOLIDS
FIGURE 20
REVERSE OSMOSIS DATA: DEPENDENCE OF SOLIDS REJECTION ON PRESSURE AND CONCENTRATION
(AMERICAN STANDARD AS-i89 MEMBRANES; CHEESE WHEY - UF PERMEATE AS FEED)
(o- 500 PSI;*- 600 PSI; D- 650 PSI;A&O- TWO TESTS AT 800 PSl)
-------
u
UJ
UJ
a:
Q
o
o
<#>
100
90 _
80
800 -
- PSI
-
0
o -
500
1
o a
D
PSI
2£
O rt
1
u ._ _0 650 PSI
— Oo0 . 600 PSI
1 1
*».
cn
n
I
o
X
Q_
Q_
Q
O
O
LU
K-
<
UJ
s:
CtL
UJ
Q_
2 0
10 —
100
FEED COD/i PPM X 10
15 0
-3
200
250
FIGURE 21
REVERSE OSMOSIS DATA: DEPENDENCE OF COD REJECTION ON PRESSURE AND CONCENTRATION
(AMERICAN STANDARD AS-ISQ MEMBRANES; CHEESE WHEY - UF PERMEATE AS FEED)
( o- 500 PSI; •- 600 PSI; a- eso PSI; A& o - TWO TESTS AT 800 PSI)
-------
1.5 _
LU
CO
o
I-
i .0
OP
LU
1-
LU
S
LU °>5
Q_
100
500 PSI
10 15
FEED % LACTOSE
2 0
2 5
dP
Z
o
H-
U
LU
~i 90
LU
LACTOSE P
D
3
°S7 V v1-' S7 800 PSI
"* " ~ 2 D-&5£_P_S1_ „
o * — — 600 PSI ' '
o
n °
0 o ° ° ° 500 PSI
)
i
i .
II l i
2 0
2 5
0 5 10 15
FEED % LACTOSE
FIGURE 22
REVERSE OSMOSIS DATA: DEPENDENCE OF LACTOSE REJECTION ON
'PRESSURE AND CONCENTRATION
(AMERICAN STANDARD AS-189 MEMBRANES; CHEESE
WHEY - UF PERMEATE AS FEED)
(o- soopsi;*- 600psi; a- esopsi; A & o
TWO TESTS AT 800 PSl)
46
-------
Dependence of Flux on Feed Concentration. Flux as a func-
tion of concentration for UF permeate is given in Fi-
gures 23r 24 and 25, for a variety of different membrane
modules. Data are shown for Universal Water Corp. modules,
Aerojet tubular membranes, Havens 300 and 310-Type modules
and American Standard modules with AS-189, AS-194 and
AS-197 membranes.
For all the modules tested, flux decreased with increasing
feed concentration as predicted by Equation (6). The osmo-
tic pressure for a 6% solids solution is approximately 100
psi. On this basis, a 24% solids solution would have an
osmotic pressure of about 400 psi. Between the range of 6
and 24% solids, osmotic pressure would increase approxi-
mately linearly with concentration. The increase in osmo-
tic pressure of the feed, therefore, accounts for the
marked reduction in flux observed for all modules. Al-
though some differences in the flux levels between modules
can be explained on the basis of difference in operating
pressure, it appears that the highest flux module was the
Universal Water Corp. module. The other membrane systems
tested had more or less the same flux, with the exception
of the AS-189 module (when new). However, comparison of
the flux data for the AS-189 membrane (when new) and flux
data at a later date shows that membrane flux decline
occurred.
Based on these data, it appears technically feasible to
produce concentrates with solids in the range of 25% from
UF permeate. Higher solids content concentrates can prob-
ably be produced at operating pressures exceeding the
level of 600-700 psi used in these experiments.
Effect of Pressure on Flux. According to Equation (6),
flux should increase with increasing operating pressure.
This is confirmed by the data shown in Figure 26, for
AS-189 membranes in the American Standard TM2-4 module.
Flux, shown as a function of feed concentration, is ob-
served to increase with operating pressure.
Effect of Circulation Rate on Membrane Performance. It is
desirable to operate at a circulation rate(through the
membrane modules) above which flux and membrane rejection
are independent of circulation rate. An experiment was
performed with a Universal Water Corp. module, American
Standard module, and Havens Type 300 module to determine
this minimum rate. All three tubular membrane systems
were used without turbulence promoters or volume displa-
cers, presently incorporated in American Standard and
47
-------
2 0
1 8
1 6
12
Q
U_
o
10
O AEROJET TUBES (590 PSIG/
25°C/ 1.8 GPM)
& UNIVERSAL WATER CORP,
MODULE NO, 349/(600 PSIG
25°C, 2 GPM)
D UNIVERSAL WATER CORP,
MODULE NO, 395.(650 PSIG
25 C/ 2 GPM)
i I i I i i I
i i i i i i I
3 it 6810
20
40 60 80 .100
% SOLIDS IN FEED
FIGURE 23
REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
FEED SOLIDS CONCENTRATION
48
-------
16
14
12
1 0
Q
CD
x 8
u_
6
4
2
0
0 HAVENS MODULE TYPE
300 (650 PSIG/ 2 GPM)
D HAVENS MODULE TYPE
\ 310 (575 PSIG/ 2 GPM)
D
O \
D\
\
\D
\
D
o
o n
\
°\
\
°\o
D
1 1 l I 1 l 1 1 1 I
3 "* 6 8 10 20 304050
% SOLIDS IN FEED
FIGURE 24
REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
FEED SOLIDS CONCENTRATION
49
-------
2 0
1 8
1 6
12
o 10
%
X
_J
U- 8
AMERICAN STANDARD
AS-189 (500 PSI/ 25°C,
WHEN NEW)
AMERICAN STANDARD
AS-189 (650 PSI/ 25°C/
AFTER 2 MONTHS LIFE)
AMERICAN
STANDARD AS-197 Xn
-(550 PSI/ 25 C) 0
i » i i i i
'AXAMERICAN STANDARD
AS-194 (575 PSI/
\U o c r )
i i i i iii
6 8 10
2 0
i*0 60 B0100
% SOLIDS IN FEED
FIGURE 25
REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON
FEED SOLIDS CONCENTRATION
-------
en
16 _
Q
12
10
600 PSIG
i o
AMERICAN STANDARD AS-189
O 600 PSIG/ 26°C/ 2 GPM CIRC,
D 800 PSIG/ 26°C/ 2 GPM CIRC,
800 PSIG
O
15
2 0
2 5
% TOTAL SOLID IN FEED
FIGURE 26
REVERSE OSMOSIS DATA: DEPENDENCE OF FLUX ON CONCENTRATION AND PRESSURE
-------
Havens modules. The data obtained indicate that for oper-
ation in 3/8" to 1/2" tubes, a circulation rate of approx-
imately 1.5 gpm is sufficient to eliminate the effects of
concentration polarization. The circulation rate was chosen
as a minimum for the pilot plant design.
Concentration of Whole Whey. Unfiltered cottage cheese
whey was concentrated to approximately 24% solids in an
experiment involving three different modules (American
Standard AS-189, Havens 300 and Universal Water Corp.).
Flux levels did not differ substantially from those ob-
served with UF permeate. On this basis, it appears that
the previous flux data can be used for design purposes for
whole whey concentration. Similarly, the rejection data
showed no major differences. This would be expected,
since the only difference between whole whey and UF perme-
ate is that the whey proteins have been removed from the
latter.
Module Life. A close record was kept of module failures
during prototype tests. This record is presented in Table
5. Shown are the name of the RO module, the date when the
module was first tested, the date when module performance
deteriorated, total hours of actual operation and the na-
ture of the failure. Eleven Aerojet tubular membranes
were tested. Although the rejection properties of the
Aerojet tubes were the best, tube failure rate was unac-
ceptable. Of the eleven tested, seven experienced rupture
at an operating pressure of 500 to 600 psi. Similarly,
Havens modules did not show the durability desired for the
pilot plant. Of four tested, two failed after relatively
short periods and two had unacceptably low rejection.
Similar performance was observed with the Universal Water
Corp. modules. Although these showed the highest flux
and excellent COD rejection, failure of both modules oc-
curred after a relatively short time. Three American Stan-
dard modules were tested, and all three performed well.
One failure occurred due to a tube rupture at a point
where a mechanical flaw existed.
It is difficult to determine the cause for module failures.
One possibility relates to the fact that for several of
the systems membranes were inserted inside tubular sup-
ports. Possibly upon startup or shutdown a partial va-
cuum could have existed within the modules causing membrane
collapse. A second potential cause of failure, involves
micro-pulsations in pressure due to the use of a piston
positive-displacement pump. Although an accumulator
52
-------
TABLE 5: REVERSE OSMOSIS LIFE DATA; PERFORMANCE OF DIFFERENT RO UNITS
Total hrs
m
Name of
RO Unit
Aerojet Tubular
Membranes
#93-290 (1.5ft2)
9B-713
03-1105
98-^581
9B-792
9B-633 "
9B-595
9B-813
Date (1)
2/10/70
2/11/70
2/11/70
2/10/70
2/11/70
3/26/70
4/7/70
4/9/70
Date (2)
3/25/70
4/2/70
4/6/70
4/9/70
4/10/70
actual use
(3)
10
62
136
106
166
160
36
30
Nature of Failure
End Seal Broke
Tube ruptured in two places
Tube ruptured in one place
Tube ruptured in one place
Tube ruptured in one place
Was good on last test date (4/22/70)
Was good on last test date (4/22/70)
Was good on last test date (4/22/70)
Aerojet New RO
Tubes
OB-3031 (1.5 ft2) 8/6/70
OB-2036 " 8/6/70
OB-2033 9/24/70
9/22/70
9/25/70
228
232
4
Tube ruptured in one place
Tube ruptured in one place
Was good on last test date (9/25/70)
Havens Modules
300 (17ft2)
210
12/16/69
1/30/70
20
No gross failure before last test
date (1/16/70), but had unsatisfac-
tory rejection
-------
TABLE 5 (Cont.)
Name of
RO Unit
510 (17ft2)
610
Universal Water
Corporation
Modules
197)
Date (1)
9/25/70
8/6/70
14 (9>5ft2) 2/4/70
Date (2)
10/2/70
8/15/70
4/22/70
Total Hours
Actual Use
(3)
80
15
395 (7ft2)
349
American Stan-
dard Modules
TM2-4 (1ft2)
TM5-14 (9 5ft2)
12/29/69 1/14/70
1/30/70 2/2/70
11/15/69
2/4/70
14
8
98
320
170
Nature of Failure
High flux and very low rejection
indicating leak
Was good on last test date 1/14/70
Was good on last test date 9/25/70
One tube ruptured due to mechani-
cal flaw
1. When the RO unit was first tested.
2. When the unit performance drastically deteriorated
3. Includes time of runs with UF permeate and cottage cheese whey
-------
effectively damped out gross pulsations, micro-pulsations
undoubtedly existed. These could have contributed espe-
cially to tube rupture.
On the basis of these limited performance data, a decision
was made to use American Standard modules in the high pres-
sure section of the pilot plant; that is, considerations
pertaining to module durability were considered most impor-
tant.
55
-------
SECTION V
DESCRIPTION OF PILOT PLANT
Following prototype experiments, a two-stage pilot plant
was designed, engineered and built by Abcor, Inc., and shipped
to Binghamton, New York, for a four-month period of operation.
This unit has since been shipped to and put in daily operation
at Crowley's Milk Co. LeFargeville facility.
ULTRAFILTRATION SECTION (LOW PRESSURE)
General:
The UF section of the pilot plant contained 210 ft2 membrane
area (192 Abcor HFA-180 tubular membranes). These were ar-
ranged in four parallel passes (modules) contained in a
stainless steel cabinet, fitted with sprays for cleaning and
sanitizing purposes. At a typical flux of 13.5 gfd approx-
imately 1,000 Ibs of cottage cheese whey could be processed
per hour.
Operation was batchwise; that is, 1 batch (or more) of whey
was charged to the system per day and concentrated without
adding additional whey. During operation, the feed was vol-
umetrically concentrated 20-fold or higher, depending on the
level of protein desired in the fluid concentrate. At the
end of the run protein concentrate was recovered from the
membrane unit by displacement with water or by draining.
Flow Schematic and System Components:
The UF section is shown in Figures 27 and 28, installed on
site in Binghamton, N.Y. Shown in Figure 27 is the stain-
less steel sanitary cabinet, and in Figure 28, a closeup of
the tubular membranes inside the cabinet. The system flow
schematic is given in Figure 29. Whey from a large storage
tank was fed gravimetrically to a smaller tank. The smaller
tank held approximately 150 gals; the larger tank held about
4,000 gals, and was filled with fresh whey on a weekly basis.
Whey was trucked from the Arkport, N.Y., cottage cheese pro-
duction facility.
Whey was pumped from the small tank into the membrane circula-
tion loop by a feed pump. High velocities over the membrane
surface were maintained by a circulation pump, in order to
maintain a high flux rate. The feed rate to the loop exceeded
the flux and some of the loop material was bled back to the
57
-------
FIGURE 27
PILOT PLANT
SANITARY CABINET HOUSING ULTRAFILTRATION MEMBRANES
FIGURE 28
PILOT PLANT
ULTRAFILTRATION MEMBRANES IN SANITARY CABINET
58
-------
AIR ELIMINATOR'
01
vo
BPR
END OF
ABCOR SYSTEM
SMALL
WHEY
STORAGE
TANK
FROM MEMBRANE
MODULES
TO MEMBRANE
MODULES
WHEY FROM LARGE
S~ORA6E TANK
'TEMPORARY LINE)
CXh-
(PS) ( ) V-4
CIRCULATION
(c s) PUMP
V-2
END OF
ABCOR SYSTEM
FIGURE 29 : ULTRAFILTRATION SECTION
-------
small tank. This had the favorable effect of partially mix-
ing the contents of the system, maximizing both flux and
protein yield. '
Two heat exchangers allowed the maintenance of a storage tem-
perature higher than that in the membrane system. This min-
imized the growth of microorganisms since the bulk of the
whey was held at a moderately high temperature during a run.
The components of the UF system and their function are des-
cribed in Table 6.
A flow schematic for the clean-in-place system provided with
the pilot plant is shown in Figure 30. This system was used
to mix and circulate cleaning and sanitizing solutions through
the membrane unit and also through the cabinet and module
sprays.
REVERSE OSMOSIS SECTION (HIGH PRESSURE)
General:
The RO section of the pilot plant contained three stages.
Each stage was operated with its own pump and control devices.
The membrane areas of each stage were 90, 60, and 60 ft2, for
stages 1, 2, and 3 respectively. Stage 1 had 18 American
Standard TM5-8 membrane modules; and stages 2 and 3 each had
12 modules. All contained RO-97 (AS-197) membranes. The
modules were mounted in an open rack (Figure 31) for conven-
ient cleaning of the exterior surfaces.
The purpose of the high pressure RO section was 2-fold:
first, to concentrate the lactose permeate from the UF sec-
tion, while reducing the effluent BOD; second, to concentrate
whole whey, again with BOD reduction.
There were two modes of operation for the high pressure sys-
tem. First, stage 1 was operated alone. With the inter-
stage tank, UF permeate or whole whey was processed batch-
wise. Alternatively, stages 1, 2, and 3 were operated sim-
ultaneously in continuous operation. Feed was taken from
the interstage tank and pumped through the 3 stage system,
continuously removing both a fluid concentrate and low BOD
permeate.
Flow Schematic/System Components: *
A flow schematic showing all major system components is pre-
sented in Figure 32. A description of each item and its
60
-------
function are given in Table 7. Figures 33 and 34 are
photographs of the high pressure system. The first photo-
graph shows the three control panels for the three high
pressure stages, and the second shows the reverse side of
the control panels.
The interstage tank is shown in Figure 27, just to the right
of the UF membrane cabinet. In addition to accumulating
permeate from the UF section for use as feed for the high
pressure section, the interstage tank was also used to mix
cleaning and/or sanitizing solutions for the high pressure
system.
The electrical control panel for both the UF and RO sections
is shown in Figure 35.
61
-------
a\
ro
RETURN FROM
BREAK POINT
DRAIN FROM
UF CABINET
HXH
V-9
400
GALLON
CIP
TANK
V-6
V-7
TO MODULE SPRAYS
TO CABINET SPRAYS
TO BREAK POINT
V-8
PUMP
7 1/2 HP
FIGURE30! CIP SYSTEM
-------
TABLE 6
ULTRAFILTRATION SECTION COMPONENTS
VI Drain valve on whey storage tank.
V2 3-way plug valve.
V3 Diaphragm valve, for control of circulation rate
in membrane system loop.
V4 Diaphragm valve, for draining system and sampling
feed stream.
PER Back pressure regulator, for control of pressure
on the low pressure side of the membrane modules.
CS Conductivity switch, for shut down of feed and
circulation pumps if air were entrained in whey
feed.
FI Flowmeter, for measure of circulation rate in
membrane loop.
PI Pressure guages.
TA Audible temperature alarm.
HE-1, Heat exchangers.
HE-2
TRC Temperature recorder/controller probe, connected
to recorder for temperature measurement; also
activated cold water solenoid valve on HE-1.
TI Temperature indicating guage in return line to
whey storage tank.
TS-1 Temperature switch, for control of hot water flow
to HE-2.
PS Low pressure switch, set at about 10 psi, for shut
down of circulation pump at low system pressure.
Air Elim. Air eliminator, for removal of air from whey in
circulation loop.
63
-------
TABLE 6 CONTINUED
ULTRAFILTRATION SECTION COMPONENTS
Feed Positive displacement pump with vented head set
Pump at 80 psi, for control of feed flow rate from
whey storage tank to membrane system, and con-
sequently - at a given flux rate through the
membranes - the recycle rate through BPR back
to storage.
Circ. Centrifugal pump, provided high circulation rate
Pump in membrane loop.
Alternate 1" pipe used for cleaning cycle.
Line
64
-------
FIGURE 31
PILOT PLANT
REVERSE OSMOSIS SECTION MODULES
FIGURE 33
PILOT PLANT
REVERSE OSMOSIS SECTION CONTROL PANELS
65
-------
INTERSTAGE TANK
DRAIN FROM LOW
PRESS. CABINET
on
en
vv-
V-3 V-4
TO V-3
BPR-2
SV-3A1 r
ALT.1 <60
STAGE 2 !
ACUM-2 [
RD-2
SV-4)
t£j &•
FI-2
V-2 FR
TO V-3
) W-3
—rxi-
BPR-3
SV-5&
STAGE 3
ACUM-3
RD-3
FIGURE 32: HIGH PRESSURE PIPING DIAGRAM
-------
TABLE 7
REVERSE OSMOSIS SECTION COMPONENTS
PI, P2, P3
ACUM/1, 2, 3
PI, P3, P5
P2, P4, P6
PS
CS
TRC
HE
SOL/1
RD/1,2,3
SV's
TA
BPR/1
BPR/2,3
Homogenizers, used as high-pressure pumps.
Pressure surge accumulators.
Pressure guages on homogenizers' outlet ports,
for measurement of pressure on high pressure
side of membrane modules.
Panel-mounted pressure guages, for measure-
ment of pressure on low pressure side of
membrane modules.
Pressure switch, for shutdown of homogenizers
if system pressure in stage 1 dropped below
60 psi.
Conductivity switch, for shutdown, of system
if air were entrained in line to suction side
of pump no. 1.
Temperature probe for recorder, actuated sol-
enoid valve controlling cooling water flow
to heat exchanger in stage 1.
Heat exchanger, for lowering temperature of
feed from interstage tank to stage 1 operating
temperature.
Solenoid valve on heat exchanger, activated
by TRC.
Rupture disks, 1,000 psi rating.
Sample valves on inlet and outlet sides of
membrane stages.
Audible temperature alarm, indicated if feed
from interstage tank were "too hot".
Stage 1 back pressure regulator, for control
of operating pressure level in stage 1.
Stage 2 and stage 3 back pressure regulators,
for increasing pressures of stages 2 and 3
differentially over that of stage 1.
67
-------
TABLE 7 CONTINUED
REVERSE OSMOSIS SECTION COMPONENTS
FI/1 Flowmeter, indicated recirculation rate in
stage 1.
VI High pressure needle valve which could be
used to isolate stage 1 from stages 2 and 3.
V2 High pressure needle valve, for shut-off of
concentrate flow.
FR Flow controller, for control of flow rate
of lactose concentrate.
TI/2 Temperature probe, indicated temperature of
concentrate leaving stage 3.
FI/2 Flowmeter, indicated flow rate of concentrate
from stage 3.
W/1,2,3 Solenoid vent valves; when system was not
operating, water was introduced through
these valves into the membrane modules to
keep them wet during shutdown.
V3 A check valve installed in the water line
to prevent fluid backing up the water line.
V4 Water regulator,- controlled water pressure to
membranes during storage.
68
-------
FIGURE 34: PILOT PLANT
REVERSE OSMOSIS SECTION CONTROL PANELS (REAR VIEW)
FIGURE 35: PILOT PLANT
ELECTRICAL CONTROL PANEL
69
-------
SECTION VI
PILOT PLANT OPERATION
GENERAL PERFORMANCE
During the period from July 22 through December 1, 1970,
the pilot plant was operated on a fairly regular basis.
Operation was generally three days per week, with a fresh
shipment of whey arriving each Monday. During this period
pilot plant performance was excellent. Capacity remained
unchanged with time, with the UF section having a capacity
of 20,000 Ibs whey/20 hour day, and the RO section having
a capacity of 10,000 Ibs whey/20 hour day. The BOD reduction
observed in the bulk of the program was from approximately
35,000 mg/1 to about 1000 mg/1 (fresh, low-acid whey). Some
exceptions were observed, in particular during periods when
one or more of the RO high pressure modules developed leaks.
In fact, the sole aspect of the pilot plant operation which
was not entirely satisfactory related to the durability of
the RO modules. At the beginning of the pilot plant opera-
tion several developed leaks and had to be replaced. In
addition, over the program period four modules experienced
tube ruptures. For the last two and a half months of opera-
tion, however, performance of the RO section was entirely
satisfactory.
The system design proved to be sanitary and produced products
with microbiological counts suitable as food or dairy pro-
ducts .
Details of the operation of the pilot plant are presented in
the sections below.
SAMPLING AND ANALYTICAL PROCEDURES
Referring to Figures 29 and 32, the following Tables list
sampling points:
71
-------
TABLE 8
SAMPLING POINTS FOR ULTRAFILTRATION SECTION
Number
Location
Purpose
Whey storage tank
Return line to small
whey storage tank
(after BPR)
Interstage tank
(feed to high pres-
sure tank)
For samples of raw whey
taken at the beginning
of a run
For samples of material
in membrane circulation
loop
For samples of UF per-
meate
TABLE 9
SAMPLING POINTS FOR REVERSE OSMOSIS SECTION
Number
Location
Purpose
1-6
SV/1 - SV/6
7-9
10
High pressure per-
meate collection
pans
Flow through FI/2
Sampling valves on in-
let and outlet sides
of high pressure mod-
ules - for samples of
feed in each of the
three circulating
stages
For samples of permeate
from each of the three
stages
For samples of final
lactose concentrate
Analyses listed in Table 10 were performed on a regular
basis.
72
-------
ANALYSES USED FOR PILOT PLANT PROGRAM
Analysis References
A
COD (Frequently calibrated
with standard glucose
solutions)
BOD
1. Standard Methods: Water
and Wastewater, 12th
V Edition, 1965.
Kjeldahl Nitrogen
Total Solids
Acidity (Lactic Acid)
Fat |
Conductivity /
2. Standard Methods: Dairy
Products, 12th Edition,
1965.
ULTRAFILTRATION SECTION
Procedure; All the experiments described involved unfiltered
cottage cheese whey. The whey was stored warm until used,
with the storage temperature ranging between 110 and 120°F.
In one experiment a synthetic curd washwater was formulated
by diluting one part of whey with four parts of cold water,
and processed.
With this storage procedure operation was with whey with an
acidity ranging between 0.5 and 1.3% (expressed as lactic
acid). This is illustrated by the data of Figure 36. Whey
was generally received with an acidity of about 0.5% (see
Aug. 10 and 17, Sept. 7, Oct. 5 and 25, and Nov. 30). In
the course of two to five days the whey acidity increased
substantially. As will be shown in the discussion of the
RO data, this increase in acidity led to an increase in the
effluent BOD. This relates to the fact that lactic acid is
.only moderately retained by RO membranes.
Referring to Figure 29, operation in a batch mode was
as follows. After a run the membrane system was cleaned
according to procedures developed in prototype experiments.
After cleaning, the system was flushed with
73
-------
dP
Q
1— «
O
JULY
AUGUST
SEPTEMBER
1970
OCTOBER
NOVEMBER
FIGURE 36
FEED WHEY ACIDITY
-------
water and stored until the next experiment. Immediately
before startup the system was sanitized with Antibac B.
Following flushing with whey, the small storage tank and the
UF system were filled with raw whey- Upon startup whey was
concentrated either for a given period of time or until a
given volume of whey had been processed. During this period
the small storage tank was kept full by the addition of
fresh whey. At a pre-selected time or whey concentration
the supply of fresh whey to the small storage tank was shut
off. At this point the system contents were concentrated
until the capacity of the small storage tank was exhausted.
This latter operation is referred to as "cook-down".
Dependence of Flux on Protein Concentration; In Figure 37
are presented data for flux as a function of the retained
protein concentration. The data presented are for runs
on four days. The runs of October 19 and 22 were performed
with whey of a high acidity. In addition, during the run
of November 30 some acidification occurred in the whey at
high protein concentrations. It appears on the basis of
these data, as well as general observations on pilot plant
performance (not discussed in detail here), that somewhat
lower fluxes were observed with high acidity whey. Also
shown in Figure 37 are the temperature profiles for the
runs.
Flux decreased with increasing retained protein concentration,
as expected. Data are comparable to prototype data which
are also shown (from Figure 13 at 120°F and 13.5 gpm).
Dependence of Flux on Time (Life Data); In Figure 38 are
shown flux data for the pilot plant during the 4% months of
operation. In the bottom plot is shown an average flux for
batch operation, to an average concentration ratio (shown in
middle plot), at an average temperature (shown in the upper
plot).
The average concentration ratio is the concentration ratio
up to the point of cook-down, and the average temperature
is the temperature during this operation. In general, scat-
ter in average flux up to cook-down can be related to varia-
tions in the concentration ratio and operating temperature.
Higher fluxes are correlatable with lower average concentration
ratios and higher operating temperatures. The most striking
observation that can be made from the data of Figure 38 is
that flux was virtually unchanged over the entire pilot plant
operation. A mean average flux for a four-fold concentration
ratio was 14 gfd at about 108°F.
75
-------
LU 120
cc
ID
I-
-------
u_130
0
^120
H
C£
111
CL
5" inn
-(MEAN) ^
i i i i i
H JULY
o
0 °
"o °°
1 1 1 1 1
AUG
0
1 ° °l
1 I 1 i°l 9
SEPT
0
°0 0 0|
1 1 ! 1 1 1 1 1 1 1 1
OCT NOV
ONCENTRATION
RATIO
o to -P
-------
In Figure 39 similar data are shown for flux during the first
and second hours of operation. Average flux during the first
hour was 19 gfd and during the second hour about 14 gfd.
These values were at an average temperature of 105°F. Al-
though more scatter exists for these data, they confirm the
conclusion that flux was unchanged during the pilot plant
operation.
Additional flux data for the entire batch concentration, in-
cluding cook-down, are shown in Table 11. Given are the date
of operation, the protein concentration obtained in the fi-
nal concentrate, the total solids content of the final con-
centrate, the percent protein on a dry solids basis, the
average flux, and the operating temperature.
TABLE 11
PILOT PLANT ULTRAFILTRATION DATA
Final
Final Protein Total
Date Concentration Solids
8/3
7/29
8/17
8/31
11/30
8.1%
6.4
14.6
8.3
9.3
14.6%
12.9
21.1
15.5
15.8
Percent Avg.
Protein Flux
(dry basis) (gfd)
. 55.5%
50.0
69.5
53.5
59.0
14.4
10.7
13.7
14.5
13.0
Avg.
Temp.
110°F
100
114
112
'114
Concentrates with protein percentages between 50 and 70% pro-
tein were generated, and average fluxes ranged between 10.7
and 14.5 gfd.
On the basis of the data for1 Figures 38 and 39 and Table 11,
it is concluded that the capacity of the UF section of the
pilot plant remained constant during pilot operation.
Dependence of Retention on Protein Concentration; At the
end of the pilot program, protein retention was determined
as a function of the protein concentration in the feed, in
a manner identical.to that discussed in "Prototype Experi-
ments: Ultrafiltration". Protein 'retention data for the
pilot plant are shown in Figure 40, along with prototype
data from Figure 6.
78
-------
u_
0
% ion
LU
DC
13
1— 110
<£
(V
UJ
^ id o
LU
H 25
2 0
Q 15
u_
L9
^
X
r>
_i
u_ 10
5
0
O
~ (MEAN) ° 0 °° o
1 , ,q , , 1 , , , p", Ltf.'S , i°J . . , • i
- (MEAN, IST HOUR) 0 ° °°o°o ° f
° 0°
D ° Q
D ^0
o °o
(MEAN/2ND HOUR) D ^ D
LJ l> LJ
D D
a a
a D
_ DO
1 I 1 1 1 1 1 1 1 1 1 1 I I I I I 1 1 1 1 I I I l 1 1 i i
JULY AUG SEPT OCT NOV
- 1970
FIGURE 39
PILOT PLANT ULTRAFILTRATION DATA:
WHEY FLUX VS THE
(15.5 GPM CIRCULATION RATE; INLET
PRESSURE 60 PSI; OUTLET PRESSURE
20 PSI; o- 1ST HOUR; ^ - 2ND HOUR)
79
-------
00
O
100
dP 90
N
z
O
*—«
I-
z
LU
»-
LU
LL)
H
O
a:
o.
8 0
7 0
FIGURE 40
PILOT PLANT ULTRAFILTRATION DATA: PROTEIN RETENTION
O PILOT PLANT, 11/30/70
PROTOTYPE DATA/ FIGURE 6
FEED PROTEIN/ %
-------
At low protein concentrations in the feed total protein re-
tention was slightly lower with the pilot unit. This is hy-
pothesized to be due to differences in the raw whey. At
higher feed protein concentrations the pilot unit gave higher
protein retentions. Retention at high protein concentration
is more directly related to true protein retention properties
of the membranes, and the pilot unit apparently had higher
true-protein retention than the prototype membranes tested.
Since these data were obtained at the end of the pilot pro-
gram and since no changes were made in the UF section during
the program, it is concluded that protein retention was
excellent throughout the pilot program.
Ultrafiltration Tests with Synthetic Curd Washwater: A
synthetic curd washwater was formulated by mixing one part
whey with four parts cold water. The temperature of the
mixture was 74°F. This temperature approximates that of
actual curd washwater. In a single experiment this mixture
was concentrated 20~fold by volume (from approximately
0.12% to 2.5% protein) with an average flux of 14.5 gfd.
The protein retention was excellent and followed the pattern
exhibited for whole whey concentration in Figure 40.
Whey washwater can be concentrated at only slightly higher
fluxes than whole whey, even though the protein concentration
is 4 to 5-fold lower. Ultrafiltration flux is not substan-
tially increased due to the fact that operation is at lower
temperatures (70-74°F instead of 125°F).
This experiment demonstrated that curd washwater can be
successfully concentrated, although to do so may not be
economically attractive. A corresponding RO test for con-
centration of the permeate is described in a later section.
REVERSE OSMOSIS SECTION
Prdcedure; All RO experiments except one were performed
with Ultrafiltration permeate. The sole exception was an
experiment involving whole whey concentration, described
separately below.
Most membrane rejection analyses involved COD assays, which
were correlated with BOD. Data for permeate samples from
the RO section exhibited a relationship, BOD = h COD.
Toward the end of the pilot program it became apparent that
the electrical conductivity of whey, UF permeates, and RO
permeates could be correlated with COD. For RO permeates
this is not surprising, since RO membranes have high salt
rejections. Any decrease in the salt rejection of the mem-
81
-------
brane would also be expected to lead to a decrease in the
rejection of organic solutes. Data showing the overall con-
ductivity-COD correlation are given in Figure 41. Over a
wide range of COD levels a general relationship appears
valid.
Operating procedure involved cleanup by flushing with line
water until the "lactose-concentrate" stream was clear. The
system was stored in this condition until the next run. Be-
fore startup the system was -sanitized with Antibac B at a
concentration of 1 oz/120 gal (10 ppm free chlorine), and
purged with;UF permeate.
Dependence of Flux and Rejection on Feed Concentration;
Flux and COD rejection were determined as a function of feed
concentration level. Data for a batch experiment (stage 1
only) are shown in Figure 42. Flux as a function of feed
concentration, expressed as COD or approximate total solids,
is shown in< the upper curve. In the lower figure COD levels
in the permeate are given; and in the middle graph, COD
rejection.
These data show a decline in flux with increasing feed con-
centration in accordance with Equation 6. In addition, COD
rejection was excellent, ranging from 98.5 to 99%. Based
on these data the mixed permeate for a four-fold volumetric
concentration would have a COD level of approximately 1250
mg/1. This; would correspond to a BOD level below 700 mg/1.
In general, somewhat higher flux levels were observed during
the pilot program. Figure 43 shows data for flux recorded
during the period from July 22, through Oct. 21, 1970. Data
points are shown for continuous operation with stages 1, 2
and 3 in three different temperature ranges. Numbers in par-
entheses near data points indicate approximate operating
pressure. From stage 1 data, it is apparent that both in-
creases in operating temperature and pressure resulted in
an increase: in flux. Similar observations are valid for
stages 2 and 3. Data from Figure 42 from operation at 84 F
and 700 psi are also shown.
The data have been scaled to an operating pressure of 750
psi in Figure 44 . The scaled factor is based solely on
the ratio o-f operating pressure; that is, fluxes are multi-
pled by (750/operating pressure). This is a conservative
scaled factor since it neglects the osmotic pressure of
the feed, which is an important consideration especially
for stages 2 and 3.
82
-------
1 0
oo
u>
a
\
CO
o
10
o
a
z
o
o
1 0
10
1 0
1 0
i o
COD
1 0
FIGURE HI
COD - CONDUCTIVITY CORRELATION FOR FEED AND PERMEATE OF REVERSE OSMOSIS SYSTEM
-------
Ll_ 00
CD
-s
^ i—< it
X CO
rj Q_
U_ O
O
100
I-
O 99
LU
LU
Q 98
O
O
Q n
o i
o o
H
UJ
LU
s
a:
LU
o_
- cr-o
1 1
o
o
o
p
1
i
i o
O
1
1 5
FEED CONCENTRATION/ % SOLIDS OR COD (MGA-X 10 4)
FIGURE 42
PILOT PLANT REVERSE OSMOSIS DATA
(8/31/70; WHEY ACIDITY = 0.8%)
84
-------
FIGURE
PILOT PLANT REVERSE OSMOSIS DATA:
DEPENDENCE OF FLUX ON CONCENTRATION
10
Q
U_
CD
STAGE 1
STAGE 2
STAGE 3
8(575K ft
o(550)
(730)
A
D D. O
A (400)
D
(730)
° (750)
x
(750)
(680) \
0 (680)
10
2 0
3 0
FEED CONCENTRATION, % SOLIDS OR COD (MG/Jl X 10"4)
O FIGURE 42; 84°F; 700 PSI
o 88 - 89°F
^ 82 - 84°F
D 78 - 80°F
85
-------
1 0
LL.
CD
X
750 PSI, 88°F
(DESIGN BASIS)
D
LINE FROM
FIGURE 42
O 87 - 89 F
A 82 - 84°F
D 78 - 80°F
0 FROM FIGURE 42
(700 RSI/ 84°F)
5 10 20 30
FEED CONCENTRATION/ % SOLIDS OR COD X 10~4(MGA)
FIGURE M
PILOT PLANT REVERSE OSMOSIS1*ATA:
DEPENDENCE OF FLUX ON FEED CONCENTRATION
(DATA FROM 7/22/70 THROUGH 10/30/70)
86
-------
8 0 '
7 0
LU
-3
UJ
ce
6 0
o
<
5 0
1. 0
Q
»—*
o
O
O
1.0
FEED % LACTIC ACID
2.0
f
o
<
0 .5
O
O
UJ
s:
or
LU
O.
.0
FEED % LACTIC ACID
FIGURE
PILOT PLANT REVERSE OSMOSIS DATA:
LACTIC ACID REJECTION
(RUN OF AUG. 14, 1.970)
87
2 . 0
-------
No trend in the data could be related to length of operating
time, and on this basis it is concluded that negligible mem-
brane flux decline occurred.
Based on these data a conservative average flux for
concentration of UF permeate from 6% to 21% solids is 5.85
gfd, and this value has been chosen as the design basis
for the demonstration plant.
Lactic Acid Rejection: Pilot data showing lactic acid re-
jection are given in Figure 45. Shown in the lower plot
is the level of lactic acid in the permeate as a function
of the lactic acid in the feed. The three points correspond
to samples from the three different membrane stages, opera-
ted in a continuous manner. From the upper plot, rejection
is observed to be in the range of 60-70%. -On the basis of
these data and prototype data it is apparent that substan-
tial lactic acid passes through RO membranes, and this con-
tributes appreciably to the level of BOD in the final efflu-
ent.
Effect of Time on Flux (Life Data): In Figures 46, 47, and
48are given water flux data for each of the three membrane
stages. These data are for two different feeds:, water and
UF permeate. Flux values for the latter were measured af-
ter the system had been filled with whey, but before substan-
tial concentration had occurred in stages 2 and 3. These
fluxes were corrected for the osmotic pressure of the whey,
assumed to be 100 psi. The correction was made according
to Equation 6.
As indicated in these three plots, leaks were observed for
all three stages on August 22. As will be demonstrated,
these leaks were related to major decreases in membrane re-
jection. On September 1, the modules in the three stages
were rearranged such that stage 1 contained only "on-spec"
modules. The remaining modules were either returned to
American Standard for new membranes (leaky modules) or
were stored (good modules). On September 22, the remem-
braned modules were returned by American Standard and re-
installed in the pilot plant. Operation from this date
proceeded smoothly, except that one module had a tube
rupture and had to be replaced.
The modules in stage 1 were used throughout the pilot pro-
gram. Life data on these modules, fot the period through
October 31, covers about 3h months of operation.
88
-------
LU
CC
LU
Q_
s:
LU
LU
9 0
8 0
7 0
c n
-------
LU
CC 80
<
cc.
LU
DL
S
LU
70
°o
D
O
n H20 AS FEED
o UF PERMEATE AS FEED
•"^
CO
Q_
O
O
10
h-
Q
LL.
CD
%
X
13
Li-
ce
LU
1-
~£-
cc 1 5
_)
CO
CO
LU
ry
Q_
O
P 10
O
CO
o
Q
LU
LU
5
O
u_
Q
LU
O
£ o
a
o
o
\^*o
0 0° X0 °
fj xs N. -^ y /*^ x.
a o " (9 oNoy u \
o HD o o o
o
—
LEAKS f 1 -
OBSERVED J L NEW MEMBRANES
i i i i i 1 i i i i i i i i i i 1 i i i i i i i i i i
^ JULY AUG SEPT OCT NOV
8 1970
FIGURE 47
PILOT PLANT REVERSE OSMOSIS DATA:
STAGE 2 FLUX
90
-------
1— •
co
Q.
O
O
^vj
1 1 1
I 1 i
LU
*" D H20'AS FEED
o UF PERMEATE AS FEED
y»— ^
LU
^3 , •
CO A 5
CO
UJ
o:
a.
o
1—
o i o
CO
o
Q
LU
LU
U_
ce s
o
u_
Q
LU
J-
O
LU
QC
oo
D 0 °
o o
o o
0 0 n
o u
o o0 o
o
._ ^>°
LEAKS t t
OBSERVED —1 L_ NEW MEMBRANES
1 1 1
^o JULY AUG SEPT OCT NOV
" 1970
FIGURE 48
PILOT PLANT REVERSE OSNOSIS DATA
STAGE 3 FLUX
91
-------
As seen in Figures 46, 47, and 48, water flux at 600 psi for
each of the stages was virtually unchanged with time, with
decreases in flux during the months of September and October
related to decreases in operating temperature. Water flux
levels were between 7.5 and 10 gfd during periods of accept-
able operation. The greater scatter for the data for stages
2 and 3 is due to inaccuracy in correcting solution flux for
the osmotic pressure of partially-concentrated UF permeate.
Although flux data for the RO section show greater variability
than data for the UF section, membrane flux for the RO modules
remained relatively constant throughout the pilot program.
Effect of Time on Rejection (Life Data); Shown in Figures 49,
50, and 51 are rejection data for each of the three stages
as a function of time. Operation began with high COD re-
jection for all three stages, followed shortly thereafter, by
major deterioration in membrane rejection efficiency. This
was related to visually observable module leaks (riboflavin
appeared in the permeate, giving a green color). However,
following the installation of new modules toward the end of
September COD rejection became excellent. This performance
continued through the end of November, at which time COD re-
jection for all three stages exceeded 97.5%.
All stages showed excellent COD rejection initially which
dropped relatively quickly- This was related to the develop-
ment of leaks in a few modules. This is demonstrated by the
data of Table 12, which give rejection data on Sept. 10 for
the stage 1 modules (see Figure 49). It is apparent that the
bulk of the modules had acceptable to excellent rejection.
One module had a gross leak; two had smaller leaks; and one
was "off-spec." Similar decreases in COD rejection for the
other stages were related to the development of leaks in in-
dividual modules.
Overall BOD and COD levels in the pilot plant effluent are
shown in Figure 52. Also given are the operating pressure
and feed whey acidity. On July 22, the first day of opera-
tion, the effluent COD was about 4000 mg/1, corresponding
to a BOD of about 2000 mg/1. This was followed by a period
during which many leaks occurred in the modules.
Continuous operation with all three stages began again toward
the end of September. After this date, COD and BOD rejection
were excellent, as shown. Variations in the oxygen demand
of the effluent were related primarily to whey acidity. It
is expected that operating pressure also played an important
role, although these data do not indicate any marked effect.
This conclusion is based, however, on the prototype data pre-
sented in Figures 20 through 22.
92
-------
100
9 0
CJ
LU
LU 80
Q
O
o
70
200
* I
Q o
O H
O
X
a
111 o?
LU \
U_ CD
—-c
Q n
O I
o o
rH
LU
I- X
<
LU o;
2: ^.
o: o
LU S
Q_
2 0
1 0
LEAK
LEAK
JULY
AUG
SEPT
1970
OCT
NOV
FIGURE
PILOT PLANT REVERSE OSMOSIS DATA:
STAGE 1 COD REJECTION
93
-------
O
LU
~~3
LU
CC.
§
100
9 0
8 0
7 0
200
Q o
O rH
o
X
Q
LU =^ 100
LU \
0^0-00 0-
o L—
J
I
2 0
Qro
O I
O O
•H
LU
I- X
<
LU a?
CC ID
LU SI
Q_
10-
JULY
U.1 .1 _.)..„ L—1_JL-.J_.
OCT NOV
FIGURE 50
PILOT PLANT REVERSE OSMOSIS DATA:
STAGE 2 COD REJECTION
94.
-------
z io°
o ;
CJ
LU
LU 80
a:
Q
O 70
c_>
200
Q o
O M
<_>
X
Q
LU o? 1 0 0
01 \
Ll_ 115
°?oo- oo-— o
1
2 0
Q 00
O I
O O
.H
LU
I- x 10
<
LU o*
51 \
a: o
LU s
o_
_ .1_L I I I I ' I I I I i I I 1 I
JULY
AUG
SEPT
1970
OCT
NOV
FIGURE 51
PILOT PLANT REVERSE OSMOSIS DATA:
STAGE 3 COD REJECTION
95
-------
TABLE 12
COD REJECTION DISTRIBUTION OF MODULES IN STAGE 1
Module #
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
ON 9/10/70
COD Rejection, % Observations
/
99.3
99.2
99.2
V
99.0 good rejection
99.0
98.9
98.8
98.7 ^
98.3 /
98.3
f
\
98.1 acceptable rejection
97.3
97.3
96.5 \
f
91.6 "OFF. SPEC.", unacceptable
86 leak,
84 leak,
50 gross leak
96
-------
•— 700
CO
Q-
UJ
o;
ra
CO 500
CO
UJ
o:
°- 400
1 . 5
OP
% 1.0
>->-
UJK
n:»-
S Q o.s
»—«
cj
< 0
/
0
UJ
Q
turn
01
>- o
XrH
O
X
UJ
UJ
s:
a:
UJ
a.
Q
UJ
X
10 I
5 H
COD
" REPLACED LEAKY MODULES
JULY
1970
FIGURE 52
PILOT PLANT REVERSE OSMOSIS DATA:
OXYGEN DEMAND REDUCTION
(RAW WHEY COD =65,000 MG/A; BOD =35,000 MG/O
(- 4-FOLD CONCENTRATION TO 22 - 25% SOLIDS)
97
-------
BOD levels, on days when low-acidity whey was used/ were in
the range of 1000 to 1500 mg/1. This corresponds to an over-
all BOD reduction of 96-97% based on a raw whey BOD of 35,000
mg/1. Reduction for the demonstration plant is expected to
be even greater since the demonstration plant will be a once-
through design, as opposed to a three-stage circulation sys-
tem. The latter design, used in the pilot plant, is less
efficient for BOD reduction, because UF permeate is processed
at higher concentrations than would be the case for a once-
through system.
Miscellaneous Reverse Osmosis Tests. (Concentration of Whole
Whey and Curd-Washwater UF Permeate); In a single experiment,
raw cottage cheese whey was concentrated in the RO section.
This test, performed on October 27, demonstrated that flux
and rejection in the high pressure section for whole whey con-
centration were comparable to values obtained-for the concen-
tration of UF permeate. Similar results have been discussed
above for prototype experiments. One marked difference be-
tween performance with whole whey and UF permeate related to
the ease of system clean-up. As will be seen in the section
on microbiological data, when UF permeate was processed in
the RO section, clean-up and sanitizing was relatively simple
and effective. However, in the experiment with whole whey
some difficulty was encountered in clean-up. Probably the
presence of residual casein fines as well as the whey pro-
teins contributed to both system fouling and the accumulation
of solids at system dead-ends. On the other hand, UF permeate
is de-proteinized and contains no suspended solids. Lactose,
salts and the residual low molecular weight organics have
high solubilities and can be removed from the system simply
by flushing with water. This difference between concentrating
whole whey and UF permeate by RO appears to be quite important,
in that some currently available RO equipment may be difficult
to clean. Processing of UF permeate alleviates this process
consideration.
When the synthetic curd washwater (20% whey in cold water)
was processed by UF, the permeate was processed in the RO
section. COD rejection in all three high-pressure stages
exceeded 98%. Fluxes were observed to be higher than for
the treatment of UF permeate from whole whey. This was as
expected, since the permeate from the synthetic curd wash-
water had a 5-fold lower solids content. Concentrating the
washwater UF-permeate 5-fold by volume produced a concentrate
with approximately 5% solids. Flux for all three high-pressure
stages was approximately 8 gfd at 72°F and 500 psi. The pro-
cessing of washwater/UF-permeate by RO presents ncr* obstacles,
and as an approximation the membrane area requirement can be
conservatively based on considering the washwater to be whole
whey.
98
-------
MICROBIOLOGICAL DATA
During the course of the pilot-plant program, samples were
analyzed for total viable bacteria and coliforms. In Table
13 are given the data for different test dates. The dates
run from the end of July through the beginning of October.
In the second column are given the dilution ratios for the
samples before culturing. The other columns give counts
for raw whey, UF concentrate, UF permeate, RO feed to stage
1, RO feed to stage 2, RO feed to stage 3, final lactose
concentrate, and the permeates from RO stages 1,2, and 3.
Several observations can be drawn. First, at no time did
the counts for either total viable bacteria or coliforms in
the protein concentrate exceed the level for grade A milk
(50,000 org/ml). This was the case even though counts in
the feed whey sometimes exceeded this level (see data for
July 31, August 18, September 29, and October 5).
Evidently, operation in the UF section at 110-120°F for
several hours did not lead to bacteriological growth. In
fact, in certain cases it appeared that decreases in plate
counts were observed. These data confirm the previously
discussed prototype data which indicated that growth of
microorganisms in cottage cheese whey is not a factor of
major importance.
In general, the UF permeate, sampled in the interstage tank,
also had low total plate counts. There were some exceptions,
however, which are attributed to improper cleaning. Although
the counts on August 14, September 29, and October 1 were
high, proper cleaning of the system resulted in a return to
low counts. (See for example, data of August 18 and October 5.)
The feed in RO stages 1, 2, and 3 was taken from the inter-
stage tank. Correspondingly, on days when the UF permeate
was contaminated (August 14, September 29, and October 1),
the feed to the RO section was also contaminated. This ex-
plains the high counts on the same dates in the RO feed sam-
ples. Additional data for October 12 and October 28 showed
contamination of the high pressure section even though the
UF permeate had low counts. This is thought to be due to
the fact that no sanitizer was used on those days.
In general, plate counts for the lactose concentrate followed
the counts for the feed to the RO system. At times, however,
counts in the lactose concentrate were higher than in the RO
system (e.g., July 29), and this is thought to be due to the
use of a Tygon tube to drain off the lactose concentrate.
99
-------
TABLE 13
BACTERIAL DATA
TOTAL PLATE COUNTS AND E. COLI COUNTS
DATE
7/29/70
7/31/70
8/10/70
8/12/70
8/14/70
8/18/70
9/2/70
9/29/70
10/1/70
10/5/70
10/12/70
10/28/70
11/7&9/70
(Water)
DILUTION
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
1/1000
1/10,000
RAW WHEY
TPC E.COLI
12,000 ,
20,000 X
41,000 ,
111,000
<30,000 ,
30,000 -1
<3,000
585,000 ,
440,000
<3,000 ,
<30,000
520,000 ,
370,000 *•
<30,000 <;,
<30,000
78,000 ,
100,000
-------
TABLE 13 Cont.
UF PERMEATE RO FEED STAGE 1 RO FEED STAGE 2 RO FEED STAGE 3
TPC
14,000
10,000
1,000
<10,000
67,000
90,000
<30,000
30,000
780,000
11,000
10,000
<3,000
<30,000
190,000
230,000
140,000
180,000
4,000
10,000
21,000
10,000
41,000
50,000
E.COLI TPC
, 4,000
<10,000
* 4,000
20,000
3
~ 36,000
40,000
650,000
, 17,000
30,000
<;L 6,000
<30,000
460,000
1 380,000
120,000
1 160,000
-, 17,000
40,00-0
,. 220,000
D 210,000
, 300,000
340,000
E.COLI TPC
<;L 8,000
<10,000
, 12,000
1 16,000
40 74'000
4U 40,000
650,000
19 46,000
*•* 20,000
-------
TABLE 13 Cont.
LACTOSE
CONCENTRATE
RO PERMEATE
STAGE 1
RO PERMEATE
STAGE 2
RO PERMEATE
STAGE 3
TPC E.COLI
46,000 ,
30,000
2,000 ,
40,000
100,000
140,000
160,000 ,
110,000
650,000
143,000 Q
70,000 JUU
<3,000
<30,000
460,000 4
240,000
110,000
140,000
48,000 ,
140,000
520,000 ,
540,000
52,000 ,
50,000
TPC
51,000
80,000
81,000
50,000
18,000
131,000
150,000
5,000
<30,000
7,000
10,000
150,000
240,000
20,000
20,000
20,000
10,000
92,000
70,000
E . COLI TPC
, 91,000
130,000
120,000
90,000
61,000
, 440,000
340,000
,-, 5,000
* <30,000
1,000
<10,OQO
<30,000
1 <30,000
8,000
10,000
, 67,000
1 30,000
, 280,000
1 290,000
E.COLI TPC E.COLI
39,000
X 20,000
, 110,000 ,
1 70,000 L
34,000
n 650,000 ,
1 700,000 -1
<1 5'°°° <1
<30,000
16,000
10,000
<30,000 .
1 <30,000
62,000 ,
60,000
520,000 ,
1 460,000
980,000 ,
1 1,400,000
<10,000
0
102
-------
The tube ran only partially full, and microbiological growth
was observed on the tube's inside surface. Proper cleaning
and sanitizing of the tube was not possible.
Microorganism counts in RO permeates are based on samples
taken from the module collection pans.- These were generally
not washed or sanitized, accounting for the relatively high
levels of microorganisms on many dates. On days when the
system was well sanitized (see for example, September 2),
however, counts in the permeates were low.
Samples from all sampling points, except for the UF concen-
trate, showed contamination at some time. It is to be noted
that proper cleaning and sanitizing was very effective in
lowering these counts. Observe the data of October 5, which
show very low counts, approaching levels for grade A milk.
Similar results are seen for September 2.
The final entry in the table gives representative data for
an experiment conducted over several days. At the end of
the run on October 28, the system was cleaned and sanitized.
At this time it was filled with water, which was sampled
over a period exceeding one week. Bacteria counts for the
system for November .7 and November 9 are shown, indicating
the presence of very few residual microorganisms. Further-
more, when the system was stored until November 30, little
bacterial growth occurred. This observation is based on the
absence of off-odors and visible colonies.
It is concluded, therefore, that not only was the system de-
sign satisfactory from the point of view of sanitation, but
also that the system could be stored for at least one month
in water without substantial microbiological contamination.
103
-------
SECTION VI
DEMONSTRATION PLANT DESIGN AND PROCESS ECONOMICS
PRELIMINARY DESIGN BASIS FOR AND DESCRIPTION OF 300,000
LBS/DAY PLANT
Design Basis for Ultrafiltration Section. The following de-
sign bases have been used for the UF section.
(a) 12 fold volumetric concentration.
(b) 20 hours operation per day in 2 batches; four
hours for cleanup and sanitizing.
(c) Operating conditions:
- 20 gpm per tube (1 inch i.d.)
- 125°F
- 50 psi inlet and 15 psi outlet pressure to
modules
(d) Average flux from pilot data of Table 11, corres-
ponding to a pilot plant throughput of 1,000 Ibs/
hr, or 13.8 gfd.
(e) Membrane area requirements:
300,000 Ibs/day:
,300
( 8
/ J..LJO
vday
/
•
)
ooox „
5
/ •"•
&
,fre
(
(
1C
24) r
20J X
hours
day
:tiona
,11
) x
1 Hpi
/ J-^ \ i
3.8 }
/ day
'operating
0 removal^
810
ft
hours
2
)
x
Description of Ultrafiltration Section. A flow schematic
for the Ultrafiltration section is shown in Figure 53. The
system contains 6 Abcor 480 ft2 modules, providing a total
membrane area of 2880 ft2, slightly exceeding the required
2810 ft2. The six modules are connected in a parallel flow
arrangement. The overall flow design is analogous to that
used in the pilot plant.
105
-------
MODULE
PERMEATE & CIP LINES
NOT SHOWN
CONCENTRATE
OUTLET <•
LEGEND
TRC - TEMPERATURE RECORDER CONTROLLER
V -3VENT VALVE
HE - HEAT EXCHANGER
PIS - PRESSURE INDICATOR SWITCH
PIS
1
t
•\
I
K
*
\
I
\ /
o^
4
(M ^ <
RETURN LINE
ONE PUMP IS
STANDBY
HE
SYSTEM
INLET
(COTTAGE
CHEESE WHEY)
SIMPLIFIED FLOW DIAGRAM: ULTRAFILTRATION SYSTEM
-------
Operation is on a batch basis. Whey is fed to a circulation
loop from a storage reservoir. Whey from the loop is bled
back to this reservoir, insuring mixing between the membrane
loop and the storage reservoir. This maximizes both system
capacity and protein yield. At the end of a batch operation,
the protein concentrate is recovered from the system by flush-
ing with water. Ultrafiltration permeate is withdrawn con-
tinuously and transferred to an interstage tank, prior to
treatment by RO.
Raw whey is fed to the circulation loop with a 180 gpm feed
pump. A second feed pump (spare) is included in the system
on continuous standby. A high circulation rate through the
modules is maintained by a 60 HP centrifugal pump. A second
circulation pump (spare) is included in the system, also on
continuous standby.
The ultrafiltration system contains a heat exchanger which
allows for either heating or cooling of the raw whey- This
permits control over operating temperature in the system.
Additional instrumentation includes:
- high/low pressure shutdown switch
- conductivity guards on pumps
- a sight glass on each module to indicate flow
- a temperature recorder/probe, to measure, control, and
record temperature :
- pressure guages to measure inlet and outlet operating
pressures (transmitted to control panel)
- process valves including two to each module (on/off
valves), controlled from operating panel
Each module is fitted with CIP sprays and piping, and the
system will include a 250 gpm spray pump, piping, and valves.
Total floor space to be occupied by the ultrafiltration unit
will be approximately 1200 ft^, in the form of a rectangle
30' x 40'.
Design Basis for Reverse Osmosis Section.
(a) Four-fold volumetric concentration of UP permeate
to 21% solids.
107
-------
(b) Twenty hours of operation per day on a continuous
once-through basis. This is followed by a four
hour cleanup and sanitizing.
(c) Operating conditions:
- 90°F
- 850 psi
(d) Average flux from pilot data of Figure 44 of 5.85
gfd.
(e) Membrane area requirements for 300,000 Ibs of whey
per day are:
(300,000) x (11/12) x (0.75) ,24. ,_1 _ ,
x (} x 1~
Description of Reverse Osmosis Section. A flow schematic of
the RO section is shown in Figure 54. The system has six
parallel passes, each containing 63 Abcor RO 13-5 modules.
These modules will have tubulence promoters and AS-197 mem-
branes. Each parallel pass will contain 850 ft2 membrane
area, corresponding to a total plant area of 5100 ft2. The
design for each pass will be for once-through operation,
with modules in a "Christmas-tree" arrangement. Several
modules will be connected in parallel at the beginning of
each pass, decreasing to a few modules at the end of each
pass.
Each pass will have its own valving arrangement so that it
can be shut down to replace any module which fails. This al
lows for membrane replacement without shutdown of the entire
plant, although isolation of any single section of the plant
will lead to a temporary reduction of capacity of 17%. Each
pass will also have its source of filtered line water fed
to the pass through check valves, permitting storage under
line water pressure.
The membrane modules will be fed by a 25 HP positive dis-
placement Quintuplex pump. A second pump will be included
as a spare, in order to avoid plant shutdown in case of pump
repairs .
Ultrafiltration permeate will be collected in a 6,QOO gallon
interstage tank, and will be pumped through the membrane mo-
dules on a once-through basis. Its temperature will be
lowered from the UF operating temperature to that of the RO
section by passage through a heat exchanger (HE) . Cooling
108
-------
water flow through the heat exchanger will be controlled by
a temperature indicator/control device (TIC). The tempera-
ture indicator will also be connected to a recorder so that
operating temperature can be monitored with time. Should
the feed flow to the RO system be interrupted, a conductiv-
ity switch (CS) will shutdown the system. An accumulator
after the high pressure pump will reduce pressure surges in
the system to acceptable levels. A safety release valve (SRV)
will vent to drain should the system be overpressurized.
Referring to Figure 54 a circulation loop will be required
around the feed pump, and someflow will be circulated through
a pressure let-down valve. Inlet and outlet pressures to
the membrane modules will be measured. A low-pressure switch
(PS) will shutdown the system in case of system failure. A
back pressure regulator (BPR) on the concentrate line controls
the system operating pressure, and the flow of the concentrate
is measured with a flow indicator (FI).
By controlling the system downstream pressure with the back
pressure regulator, and the recycle rate on the feed pump,
the inlet pressure and the concentrate flow will be auto-
matically fixed. If the amount of recycle on the feed pump
is decreased, the inlet pressure will rise as will the lac-
tose concentrate flow rate.
Permeate from the high pressure modules will be manifolded
and discharged to the drain by gravity, or to a reservoir
for reuse within the dairy.
Floor space requirement for the RO section will be approxi-
mately 750 ft2, in the form of a 20' x 30' rectangle.
CAPITAL COSTS
Capital cost items are given in Table 14. Costs for the
ultrafiltration and reverse osmosis sections (installed) are
$217,000 and $360,000 respectively. Five tanks will be
required. These include four 20,000 gallon tanks for raw
whey and water storage. One 6,000 gallon surge tank will be
needed for the UF permeate (feed tank for the RO section).
Total costs for the tanks, including piping and valves, is
projected to be $50,000.
Additional capital costs include $40,000 for installation,
and $30,000 for a building to house the plant. Total capital
costs for the plant are projected to be $697,000.
109
-------
LEGEND
CS - CONDUCTIVITY SWITCH
HE - HEAT EXCHANGER
TRC - TEMPERATURE RECORDER CONTROLLER
ACCUM - ACCUMULATOR
PI - PRESSURE INDICATOR
SRV - SAFETY RELIEF VALVE
FI - FLOW INDICATOR
PS - PRESSURE SWITCH
BPR - BACK PRESSURE REGULATOR
6000
•GAL
TANK
UF
PERMEATE
ALTERNATE
LIN.ES
PUMP
(STANDBY)
-CXh—
z*
•00-
iXl-
FILTERED
WATER
MODULE
MODULE
MODULE
MODULE
MODULE
L
-CXh-
HXH
HXJ-
MODULE MX3-
LACTOSE
CONCENTRATE
rFI
-Nl—I
BPR
PERMEATE FOR REUSE
OR TO DRAIN
FIGURE 54
SIMPLIFIED FLOW DIAGRAM: REVERSE OSMOSIS SYSTEM
-------
TABLE 14
PROJECTED CAPITAL COSTS FOR 300,000 LBS. WHEY/DAY PLANT;
Equipment
Ultrafiltration Section (CPI tanks included),
installed $217,000
Reverse Osmosis Section, installed 360,000
Tanks '
4 - 20,000 gals, plastic silos for whey
and water storage
1 - 6,000 gals. UF permeate surge tank
Total for 5 tanks with piping and
valves 50,000
627,000
Installation Costs (other than membrane 40,000
equipment)
Building Costs (3000 ft2) 30,000
TOTAL CAPITAL COSTS $697,000
111
-------
OPERATING COSTS
Operating costs are shown in Table 15. Among the items
listed, costs for power and cooling water are based on actu-
al operating data from the pilot plant. The membrane equip-
ment itself, has been depreciated on two bases. Membrane
modules, less the annual charge for membrane replacement,
have been depreciated over five years. The associated mem-
brane hardware (pumps, valves, controls, etc.) has been
depreciated over ten years, as have been the remaining capi-
tal cost items.
The annual operating costs will be approximately $220,000.
PROJECTED BY-PRODUCT VALUES AND PROCESS PROFITABILITY
Fractions obtained from the membrane process have useable
characteristics. In treating 96,000,000 Ibs of cottage
cheese whey per year, 576,000 Ibs of protein (at 0.6% in
whey) and"4,320,000 Ibs of lactose (at 4.5% in whey) will
be recovered. The value for these by-products when used
as food ingredients will be substantial in dollars. The net
annual profit, the difference between income and operating
expenses, is expected to show a highly favorable return on
investment.
COSTS FOR OTHER COMMERCIAL PLANTS
It should be noted that the capital costs projected here
are for the first combined UF and RO plant designed speci-
fically for use in a dairy. The costs" of subsequent plants
may be significantly lower. Operating costs would also be
expected to be reduced. The construction and operation of
the full-scale plant in Phase II of the project will provide
more accurate cost information required for widespread ac-
ceptance of UF and RO.
112
-------
TABLE 15
PROJECTED OPERATING COSTS FOR 300,000 LBS
WHEY/DAY PLANT
Basis: 320 days/yr or 7680 hrs/yr
Labor and Overhead:
Supervisory:3 hrs/day x 320 days/yr $ 4,800
@ $5.00/hr
General Utility:7,680 hrs x $3.00/hr 23,040
Lab Tester:2 hrs/day x 320 days/yr 1,920
@ $3.00/hr
Mechanic: 1 hr/day x 320 days/yr 1,280
@ $4.00/hr $ 31,040
Fringe and overhead @ 40% of wages 12,400
Total Labor $ 43,440
Annual Membrane Replacement Cost @ $5/ft2 39,900
*Power @ 1.2*/kwh 5,000
*Cooling Water 4,000
Steam 3,000
Cleaning Chemicals ($10/day) 3,200
Disposables (e.g., pump gaskets, seals etc) 3,200
Professional Services (Outside maintenance costs, 10,000
engineering services, consultants)
Interest and Taxes @ 3% of Capital 20,910
Depreciation:
Membrane Modules ex membranes (§20% 42,700
Membrane Process Hardware (pumps, valves, 32,400
controls, etc.) @ 10%
Remaining Capital Items @ 10% 12,000
Projected Annual Operating Costs $219,750
*Actual costs based on scale-up of pilot plant utility usage.
113
-------
SECTION VII
INDUSTRY INTEREST (INCLUDING PUBLICATIONS)
The Crowley's demonstration project will be a success for
EPA only if cottage (and other) cheese makers throughout
the country accept ultrafiltration and reverse osmosis as
whey treatment processes in their own plants.
Measures of industry acceptance are the papers and publi-
cations reporting on the project (Table 16) and the number
of visitors to the pilot plant facility (Table 17).
115
-------
TABLE 16
PUBLICATIONS AND PAPERS
1. R.R. Zall, talk given at Metropolitan Dairy Technology
Society, New York, N.Y., (March, 1970).
2. B.S. Horton, et. al. , "Membrane Separation Processes
for the Abatement of Pollution from Cottage Cheese Whey" ,
presented at the Cottage Cheese and Cultured Milk Pro-
ducts Symposium, University of Maryland (March 11, 1970).
3. B.S. Horton, "Prevents Whey Pollution-Recovers Profit-
able Byproducts", Food Engineering, Vol. 42 , No. 7, pp.
81-83 (July, 1970) .
4. B.S. Horton, et. al., "New Method for Economical Control
of Pollution Caused by Cheese Wheys", presented at SOS/
70, Washington, D.C. , (August 14, 1970).
5. R.L. Goldsmith, et. al . , "Recovery of Cheese Whey Pro-
teins through Ultraf iltration" , presented at SOS/70
Washington, D.C. (August 14, 1970).
6. R.R. Zall, talk given at American Cultured Diary Pro-
ducts Institute, Cornell University, Ithaca, N.Y.
(September, 1970) .
7. B.S. Horton, "Ultraf iltration and Reverse Osmosis for
Processing Cottage Cheese Whey", presented at American
Cultured Dairy Products Institute Annual Meeting, Cornell
University, Ithaca, New York (September 9.- 1970).
8. B.S. Horton, "Ultraf iltration and Reverse Osmosis for
Processing Cheese Wheys", presented at South Dakota
State Dairy Association Annual Convention, Sioux Falls,
S. Dakota (September 22, 1970).
9. B.S. Horton, et. al., "Membrane Separation Processes
for the Abatement of Pollution from Whey", XVIII Inter-
national Dairy Congress, Sydney, Australia (October 12-
16, 1970) .
'&
10. B.S. Horton, talk given for the Massachusetts Institute
of Dairy Science, Boston, Mass. (November, 1970).
116
-------
11. Goldsmith, R.L. et. al., "Industrial Ultrafiltration",
presented at American Chemical Society Symposium on
Membrane Processes, Chicago, Illinois, (September 13-
18, 1970) .
12. Goldsmith, R.L., et. al., "Membrane Processing of
Cottage Cheese Whey for Pollution Abatement", presented
at Second National Symposium on Food Processing Wastes,
Denver, Colorado, (March 23-26, 1971).
13. Horton, B.S., talk given at Quality Chek'd Products
Annual Production Meeting, Memphis, Tennessee, (March,
1971).
14. Goldsmith, R.L., et.al., "Industrial Ultrafiltration"
in Membrane Processes in Industry and Biomedicine,
Plenum Press, 1971, Pages 267-300.
15. Horton, B.S., et.al., "Membrane Separation Processes
for Treatment of Whey and Rinse Waters in Cottage
Cheese Plants of All Sizes", 1971 Conference of American
Cultured Dairy Products Institute, Athens, Georgia,
(September 8-9, 1971).
16. Horton, B.S., talk given at Nordica Foods Annual Seminar,
Sioux Falls, South Dakota, (October, 1971).
17- Zall, R.R., et.al., "Membrane Processing of 300,000
Ibs/day Cottage Cheese Whey for Pollution Abatement,
Phase II", presented at AIChE 64th Annual Meeting,
San Francisco, California (November 28-December 2, 1971).
18. Horton, B.S., et.al., "Whey Processing with Sanitary,
CIP Membrane Equipment", presented at 1971 Winter
Meeting, ASAE, Chicago, Illinois, (December 7-10, 1971).
117
-------
TABLE 17
VISITORS TO WHEY PILOT PLANT AT CROWLEY'S MILK CO.
NAME & TITLE
COMPANY & ADDRESS
DATE
VISITED
T.W. Brooks, Ph.D.
Section Leader
Professor James Harper
Mr. J.L. Maubois
Mr. J.P. Barbier
James Garrison, V.P. Mfg,
Delbert Harmon
Dr. Larry Claypool V.P.
Research
Dick Bonney Mgr. Mfg.
Operations
Burdet Heinemann, Pres.
Dr. Curtis Hallstrom
Charles Hungerford
Russell W. Carnahan
Marty Engel
Eugene Kolen
Hubert Fatta
Director
Calgon Corp.
Calgon Center
P.O. Box 1346
Pittsburgh, Pa.
15230
Ohio State University
Columbus, Ohio
Institut National de la
Recherche Agronomique
Laboratoire de Recherches de
Technologie Laitiere
65, Rue de Saint-Brieuc
35 Rennes, France
Mid-America Dairymen, Inc.
P.O. Box 1837
S.S. Station
Springfield, Mo. 65805
10/5/70
General Mills
Central Research Lab.
James Ford Bell Res. Center
9000 Plymouth Avenue
Minneapolis, Minn. 55427
Kraftco Corp.
Research and Dev. Div.
801 Waukegan Road
Glenview, 111. 60025
Kraftco Corp. (see above)
Cooperative du Val d'Or
45, Saint-Aignan-des-Gues
Loire-et-Ch. France
10/14/70
10/20/70
10/19/70
10/19/70
10/27/70
10/28/70
10/28/70
118
-------
NAME & TITLE
COMPANY & ADDRESS
DATE
VISITED
L. Benghouzi
Ingenieur
Lloyd J. Peterman
Edward S.K. Chian, Sc.D.
Mgr. of Technology
William 0. Miller
Anton Amon
K.J. Kirkpatrick
Avelino Salgado de
Oliveira, Jr.
Engineer
Victor Moreno, Ph.D.
R.E. Farrar
Exec. V.P.
Steven B. Chall
Sr. Engineer
Arno Hus te
Sr. Research Specialist
George Rey
Direction Departmentale 10/28/70
de 1'Agriculture du Loiret
21, rue Eugene-Vignat
45 Orleans, France
A&P National Dairy Div. 9/8/70
102 Revere Drive
Manitowoc, Wise. 54220
American Standard Conseps 8/21/70
P.O. Box 5000
Hightstown, N.J. 08520
(Abcor)
Coca Cola (see below) 8/19/70
Coca Cola Export Corp. . 8/19/70
515 Madison Avenue
New York, N.Y. 10022
New Zealand Dairy Research 9/19/70
Institute
P.O. Box 1204
Palmerston North, N.Z.
Lacticinios bom Pastor, LDA
Rua Carlos Mardel 38-C
Lisboa-1 Portugal
General Foods 9/29/70
Tarrytown, N.Y.
Dairy Research Inc. 9/30/70
120 Eastman Building
Arlington Heights, 111.
60004
General Foods 9/29/70
Technical Center
Tarrytown, N.Y. 10519
General Foods (see above) 9/29/70
E.P.A. 9/17/70
Alexandria, Va.
119
-------
NAME & TITLE
COMPANY & ADDRESS
DATE
VISITED
Bob Burm
E.T. Wilhelm
Manager
Charles Zanzig
Dave Hennigh
Dept. Manager
Professor Richard March
George Walgrove,
Plant Sup.
Walter Kneeland
Proj. Leader
Gerald W. Smith
Gen. Manager
John B. Townsend
H.N. Jensen
Assistant Director
Stephen J. Wolff
Allyn Richardson
A. de la Bourdonnaye
Ingenieur
E.P.A. 9/23/70
Central Soya 10/7/70
Process & Product Dvlmt.
1825 N. Laramie Avenue
Chicago, 111.
Foremost Foods Co., R&D 10/10/70
6363 Clark Avenue
Dublin, California 94566
Safeway Stores 11/9/70
2538 Telegraph Avenue-
Oc-kland, California 94612
Cornell University 10/30/70
Ithaca, New York
H.P. Hood & Sons 11/10/70
500 Rutherford Avenue
Boston, Mass. 02129
Plant at St. Albans, Vt.
Meyer-Blanke Co. 11/30/70
5432 Highland Drive
St. Louis, Missouri 63110
Dart Industries 11/23/70
P.O. Box 3157
Terminal Annex
Los Angeles, California 90054
Pevely Milk Company 11/30/70
1001 South Grand Blvd.
St. Louis, Missouri 63104
E.P.A. 11/30/70
N.E. Region
JFK Federal Building
Government Center
Boston, Mass. 02203
Ministere de 1'Agriculture 11/70
Dir. des Industries Agricoles
et Alimentaires
3, Rue Barbet-de Jouy
Paris Vile France
120
-------
SECTION IX
ACKNOWLEDGEMENTS
The support of the project by the Environmental Protection
Agency and the help provided by Mr. William Lacy, Mr. George
Keeler, and Mr. Allyn Richardson, the Federal Grant Project
Officer, is acknowledged with sincere thanks. The project
directorship was under the responsibility of Robert R. Zall
of Crowley's Milk Co.
Dr. Robert L. Goldsmith of Abcor, Inc. held program respon-
sibility for Abcor,- Inc's. subcontracted activities, and
prepared this report with the aid of Mr. Bernard S. Horton
(Abcor). Messrs. Sohrab Hossain (Abcor) and Michael Tan
(Abcor) executed the prototype experimental program.
Dr. David J. Goldstein (Abcor) was in large part responsible
for the design and construction of the pilot plant. Many
helpful discussions were held with Drs. Robert S. Timmins
and William Eykamp (Abcor), and Professors Steven R. Tannenbaum
and Tony Sinskey (M.I.T., Dept. of Nutrition and Food Science).
Mr. Patrick Crowley at Crowley's Milk Co., aided in the
operation of the pilot plant. Much of the analytical work
at Crowley's Milk Co. was performed by Mrs. Anna Wichtowska,
Research Associate. Substantial assistance and support for
the project were provided by Mr. William Burtis of Crowley's,
the project finance officer, and also Messrs. Vincent and
Elmer Crowley.
Membrane equipment formerly fabricated by the ConSeps Depart-
ment of American Standard Corp. is presently manufactured by
Abcor, Inc. All membrane equipment for the Phase II, 300,000
Ibs/day plant will be manufactured by Abcor, Inc.
121
-------
SECTION X
REFERENCES
1. Anonymous, Volume III, Industrial Waste Profiles, No.
9-Dairies, "The Cost of Clean Water", FWPCA, Washington,
B.C., September, 1967.
2. McDonough, F.E., "Whey Concentration by Reverse Osmosis",
Food Engineering, Vol. 40, No. 3, March, 1968.
3. Marshall, P.G., Dunkley, W.L., Lower, E., "Fractionation
and Concentration of Whey by Reverse Osmosis", Food
Technology, Vol. 22, No. 8, pp. 37-44, August, 1968.
4. Anonymous, "USDA Studies Reverse Osmosis as Whey Disposal
Method", New Release No. USDA 1396-68, Washington, D.C.,
May 1, 1968.
5. deFilippi, R.P., Goldsmith, R.L., "Application and
Theory of Membrane Processes for Biological and Other
Macromolecular Solutions", Membrane Science and Tech-
nology, ed. , James E. Flinn, Plenum Press, 1969".
123
-------
1
Access/on Number
w
5
2
Organization
Crowlev ' s
Subject Field & Croup
05D
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Milk C omn an v
Binghamton, New York
Title
Membrane Processing of Cottage Cheese Whey for Pollution
Abatement
10
Authors)
Goldsmith, R.L.
Horton, B.S.
of Abcor, Inc.
Cambridge, Mass.
16
Project Designation
EPA, WQO Project 12060 DXF
Note
22
Citation
23
Descriptors (Starred First)
*Cottage Cheese Whey Treatment, Cottage Cheese Processing, Whey,
Food Processing, Membrane Separation, Pollution Abatement, Reverse
Osmosis, Ultrafiltration
25
Identifiers (Starred First)
^Cottage Cheese Whey, *Whey, Whey Treatment, Dairy Waste, Waste Treatment,
27 Abstract A two-step membrane process was demonstrated for treatment of
Cottage cheese whey. The process produces valuable protein and lactose by-
products while reducing BOD of effluent whey. The process was studied in
detailed prototype experiments at Abcor, Inc., Cambridge, Mass., and a 10,000
Ibs/day pilot plant at Crowley's Milk Company, Binghamton, New York.
In the two-step process, a protein concentrate is first recovered by ultra-
filtration. In the second step, ultrafiltration permeate (de-proteinized
whey) is concentrated by reverse osmosis providing a lactose concentrate. The
protein concentrate can be further concentrated and/or dried; the lactose
concentrate can be further concentrated and lactose recovered by crystalliza-
tion, or otherwise processed. Alternatively, the membrane concentrates are
usable as fluid products.
Operation of the pilot plant was successful and almost troublefree. BOD
reduction of raw whey was about 97%, from initial values of about 35,000 to
less than 1,000 mg/1. Membrane life was excellent, and membrane fluxes were
economically high.
The pilot plant produced protein and lactose products with low total plate
and nil coliform counts. Using the cleaning procedure developed in prototype
tests, total plate counts were typically below 50,000 org/ml.
Abstractor
. Robert Go!
d smith . . .
Institution
Abcor ,
Inc. ,
Cambridge r
Massachusetts
WR:I02 (REV. JULY 1989)
WRSIC
SEND. WITH COPY OF DOCUMENT. TOl WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON. D. C. 20240
«J.S. GOVERNMENT PRINTING OFFICE: 1972-484-483/94 1.3
------- |