You are on page 1of 11

Reverse Osmosis of Dairy Liquids

J. H I D D I N K , R. de BOER,
and P.F.C. NOOY
Netherlands Institute for Dairy Research (NIZO)
P. O. Box 20
6710BA Ede, The Netherlands

ABSTRACT factory is savings on costs of transport from the


Permeate flux during reverse osmosis cheese factory to a central whey processing
of skim milk, Gouda cheese whey at plant.
various pH, desalted whey, ultrafiltration At the moment, seven RO plants with a
permeate, and lactose solutions were water removal capacity ranging from 3.5 to 15
studied. The flux-limiting factors of the m3/h have been installed, four of them using
feed appeared to be osmotic pressure and tubular membranes and three of them using flat
fouling agents. Protein in Gouda whey at membranes. In principle, two processes can be
pH 6.6 did not give rise to fouling; distinguished. One process involves concentra-
however, after desalting or decrease of pH tion of whey at 30 C in a tubular single-pass
to 4.6, fouling by whey protein occurred. system, to a concentration ratio of about 2:1.
If Gouda whey was concentrated at 30 C In the second process whey is cooled to 10 or
over a concentration ratio of 1.6:1, 17 C and then concentrated, batchwise or in a
calcium-phosphate precipitation caused continuous system, to a concentration ratio of
strong fouling of the membrane. This 2:1 or 3:1. An important factor for RO is the
fouling can be prevented by removing permeate flux, this factor governing to a large
calcium from the whey or by acidifying extent the economy of the process. The factors
to a pH of about 6.0. During reverse affecting the permeate flux, in addition to the
osmosis of skim milk, casein is the properties of the membrane, are the composition
principal fouling material. of the feed and the process conditions.
To understand better the influence of Though RO membranes can be made of a
the process conditions on the permeate variety of materials (cellulose acetate, poly-
flux, the effect of concentration polari- amide, cellulose nitrate, sulfonated poly-
zation and fouling was calculated for phenylene oxide, polybenzimidazolone), up to
various conditions. It appeared possible now only cellulose acetate (CA) has proved to
to achieve a reasonably low concentration be commercially acceptable for RO of food and
polarization and fouling by applying, in a dairy liquids. For process conditions, tempera-
tubular system, flow velocities of 2 m/s ture is of particular importance. Raising the
for whey and of 2.5 m/s for skim milk. temperature of the feed increases the permeate
Furthermore, for skim milk an optimum flux, but to work under completely safe
process pressure was about 3.5 MPa. bacteriological conditions a temperature below
10 C or above 50 C should be used. Cellulose
INTRODUCTION acetate (CA) membranes should not be operated
In the dairy industry reverse osmosis (RO) is above 35 C since these membranes are subject
used more and more as a concentration tech- to hydrolysis and compaction at higher tempera-
nique (6, 11, 13, 26). The main advantage of tures. This consideration should lead to a
RO over evaporation is the low energy consump- process temperature of 10 C. In practice,
tion (2). In the Netherlands RO has been however, temperatures of 30 C (temperature of
in use for some years for the concentration of the Gouda cheese whey) and 17 C (temperature
whey and ultrafiltration permeate. The main of the whey after cooling with cold milk in a
reason whey is concentrated at the cheese heat exchanger) are also used. At these tempera-
tures extensive precautions, such as a short
residence time and intermediate disinfection,
must be taken to prevent bacteriological
Received April 2, 1979, spoilage.

1980 J Dairy Sci 63:204--214 204


REVERSE OSMOSIS OF DAIRY LIQUIDS 205

Concerning the properties of the feed, understanding of the effect of both composition
osmotic pressure and the fouling tendency must of the feed and process conditions (temperature,
be considered. The osmotic pressure of dairy flow velocity, and pressure) on the permeate
liquids is determined mainly by the low molecu- flux. In particular we tried to quantify the
lar weight solutes such as salts and lactose. It effect of concentration polarization and fouling.
must be overcome by the applied process Our study was restricted to skim milk, Gouda
pressure. cheese whey, and their derivatives.
Fouling can be caused by precipitation of
certain salts and by build up of a deposit of MATERIALS AND METHODS
colloidal material. It should be prevented by a
Experimental Procedure
proper pretreatment of the feed and by applying
proper process conditions. A number of investi- A laboratory recirculation RO plant was
gations referred to in the literature deal with used, consisting of two modules type B1
the fouling properties of the various components (length 3.66 m) from Paterson Candy Inter-
of whey and skim milk and with the composition national Ltd., England. The modules were
of the deposit. Lira et al. (15), Peri and Dunkley provided with tubular CA membranes type
(18), and Mehta (16) concluded that during T2/15W with an inner diameter of 1.3 cm and a
concentration of cottage cheese whey, the salt rejection of about 93%. The membrane area
principal fouling material is protein, and per module amounted to 2.55 m 2.
particularly casein. The selective concentration The experiments were by batch; from a tank
of casein was explained by its low diffusion the liquid was pumped through the system by a
coefficient. Smith and McBean (28)investigated piston pump and returned to the tank while the
membrane fouling in RO of cheddar cheese permeate was drained. Each experiment took 2
whey and hydrochloric acid casein whey. They to 4 h. The process temperature was maintained
found that calcium plays an important part in at 10 or 30 C. If a process temperature of 30 C
membrane fouling, in particular precipitation of was used, the liquid was heated in the tank
insoluble calcium salts. The pretreatments before starting the experiment. Pressures were
developed by Hayes et al. (12) for reduction of between 2.5 and 4.5 MPa and flow velocities
fouling in uhrafiltration led to increased fouling between 1 and 2.6 m/s. The flow velocity was
in RO. This was ascribed to the higher ionic calculated from the rate of flow divided by the
concentration that with RO takes place in the area of cross-section of the membrane tube.
surface layer on the membrane. Glover and Before use, the skim milk and the whey were
Brooker (10) and Skudder et al. (27) studied by separated (6800 × g) and pasteurized at 72
electron micrographs the structure of deposits C/15 s.
formed during RO of whole milk. The deposit
consisted mainly of casein micelles while Dairy Liquids
Ca-phosphate was precipitated in the deposit Composition of the dairy liquids tested is in
during concentration. Table 1. When whey was desalted, a strong acid
Concerning the process conditions which cation exchange resin and a weak base anion
should be applied for RO of whey and skim exchange resin were used. About 95% of the
milk much work has been done. The essential mineral matter was removed. The ultrafiltration
difference between the early work (8, 15, 18) permeate (UF-permeate) was obtained by UF
and more recent work (5, 25, 26, 27) is the of whey at 10 C. The pH of all liquids referred
difference in applied flow velocity. In the early to in Table 1 was 6.5 to 6.7 or was adjusted to
work the flow velocity was .10 to .55 m/s (at a that pH. In some of our experiments we used
tube diameter of 1.25 cm), but in the more Gouda cheese whey acidified with HCI to pH
recent work flow velocities of 2 to 3 m/s were 6.0 or 4.6.
applied at a similar tube diameter. With these In some experiments whey was used from
higher velocities much better permeate fluxes which 90% or more of the calcium was removed
were achieved. Typical applied pressures in by ion exchange with a weak cation exchange
both cases were 3 to 5 MPa and temperatures resin, which was regenerated with HC1 and
between 10 and 35 C. NaOH. After regeneration the resin is in Na +
Our work was aimed at acquiring a better form. In Table 1 the osmotic pressure for the

Journal of Dairy Science Vol. 63, No. 2, 1980


206 HIDDINK ET AL.

TABLE 1. The composition, the osmotic pressure at 10 C and the dynamic viscosity at 10 C of the dairy liquids
used.

Total Whey Osmotic Dynamic


solids Lactose Ash protein Casein pressure viscosity

(%) (MPa) (mPa.s)


Skim milk 8.9 4.6 .7 .7 2.8 .72 2.8
Gouda whey pH 6.6 5.5 4.1 .5 .7 .. .61 1.6
Desalted whey pl-I 6.6 5.1 4.1 .03 .6 .. .35 1.6
UF-permeatc 5.0 4.1 .5 .2 a .. .61 1.5
Lactose solution 4.0 4.0 . . . . . . . . .30 1.4

aNonprotein nitrogen × 6.38.

various liquids is given also. T h e o s m o t i c also. T h e increasing o s m o t i c pressure o f t h e


pressure was d e t e r m i n e d b y m e a s u r i n g t h e r e t e n t a t e caused a decreasing driving f o r c e for
freezing p o i n t d e p r e s s i o n w i t h a K n a u e r l lalb- w a t e r p e r m e a t i o n (see A p p e n d i x A, e q u a t i o n
m i k r o - O s m o m e t e r t y p e M. T h e o s m o t i c pres- [1] ). T h e d i f f e r e n c e b e t w e e n t h e p u r e w a t e r
sures at 10 C a n d 30 C were c a l c u l a t e d f r o m t h e flux and t h e calculated curve A in Figure 1
freezing p o i n t d e p r e s s i o n a c c o r d i n g t o t h e r e p r e s e n t s t h e effect of t h e r e d u c t i o n o f t h e
m e t h o d d e s c r i b e d b y Reid (21). driving force b y t h e o s m o t i c pressure.
T h e viscosity o f t h e various liquids was Curve A w o u l d be f o u n d if we h a d to cope
d e t e r m i n e d b y an E M I L A R h e o m e t e r ( T a b l e 1). o n l y w i t h an e f f e c t o f increasing o s m o t i c
Samples were a n a l y z e d a c c o r d i n g to t h e pressure. Curve B in Figure 1 w o u l d be f o u n d if
f o l l o w i n g m e t h o d s : t o t a l solids b y d r y i n g in an t h e effect o f c o n c e n t r a t i o n p o l a r i z a t i o n also is
oven at 105 C; t o t a l p r o t e i n b y t h e Kjeldahl t a k e n into a c c o u n t . C o n c e n t r a t i o n p o l a r i z a t i o n
p r o c e d u r e ( p r o t e i n f a c t o r 6.38); lactose a c c o r d - is d e f i n e d as t h e b u i l d - u p o f s o l u t e s n e a r t h e
ing t o L u f f - S c h o o r l ( 2 4 ) ; ash b y d r y i n g in m e m b r a n e , w h i c h as a result causes an increase
an o v e n at 105 C a n d h e a t i n g in an electric
m u f f l e f u r n a c e at 550 C for at least 8 h; a n d
calcium b y a f l a m e p h o t o m e t e r .
9Pr'leate fix ,1121}

~_ pure waler flux


31/ . . . . . . . . . . . . %. . . . . . . . . . . . . . . . .
RESULTS A N D DISCUSSION

Influence of Composition 1

of Feed at 10 C \ effect o~ (on{enlrahon p31arh'dtl )n

A series o f R O e x p e r i m e n t s was w i t h liquids \


\
in T a b l e 1. The liquids were c o n c e n t r a t e d
d u r i n g t h e e x p e r i m e n t s a t a process t e m p e r a t u r e
o f 10 C, an average pressure of 4 MPa, a n d a
flow v e l o c i t y o f 2 m/s. A t this flow v e l o c i t y t h e 5
x• ~
e e 0 0 n9 f ~
tollllng
~a ~deseltedwhey~ "~
R e y n o l d s n u m b e r s were 5 0 0 0 to 2 0 0 0 0 ,
d e p e n d i n g o n t h e viscosity o f t h e liquid. n Z 4 6 8 10 1.2 I4 16 18 20 22 2.4 20
0sm0hc pressure ,~Pal
The experimentally determined permeate
Figure I. Permeate f l u x versus osmotic pressure o f
fluxes are in Figure 1 as a f u n c t i o n o f t h e the b u l k liquid f o r RO o f various liquids. The experi-
o s m o t i c pressure o f t h e b u l k liquid. During mental fluxes are compared with those calculated
c o n c e n t r a t i o n t h e o s m o t i c pressure o f t h e assuming no concentration polarization and no f o u l i n g
r e t e n t a t e increased, a p p r o x i m a t e l y in p r o p o r t i o n (curve A) and with those calculated assuming o n l y
concentration polarization (curve B). Process condi-
w i t h t h e c o n c e n t r a t i o n ratio, w h i l e t h e p e r m e a t e
tions: T = 10 C, P = 4 MPa, v = 2 m/s. • lactose,
f l u x decreased. In Figure 1 t h e p u r e w a t e r flux o UF-permeate, X sweet whey, o desalted whey, and
u n d e r t h e r e l e v a n t process c o n d i t i o n s is i n d i c a t e d skim milk.

Journal of Dairy Science Vol. 63, No. 2, 1980


REVERSE OSMOSIS OF DAIRY LIQUIDS 207

of osmotic pressure at the membrane surface strong flux decline was not found, and the
(see Appendix A). Curve B was achieved by the permeate flux resembled that of the calculated
evaluation of equation [6] of Appendix A. In curve B. Based on this information it was
Figure 1 the experimental permeate fluxes for assumed that during the RO of whey and
RO of the lactose solution, the UF-permeate, UF-permeate, precipitation of Ca-salts on the
and the whey resemble more or less the calcu- membrane occurred. Solubility of Ca-phosphate
lated curve B. The experimental permeate decreases with increasing temperature (1, 20,
fluxes for desalted whey are lower than those 23). Apparently during the RO of whey and
calculated for only concentration polarization. UF-permeate at 30 C at a concentration ratio
Here fouling also plays a role. Fouling is defined of about 1.6:1 fouling of the RO membrane by
as the deposition of colloidal material on the precipitated Ca-phosphate started. This fact
membrane surface, giving rise to an additional agrees with observations of Roger et al. (22)
hydraulic resistance, acting in series with the that in UF-permeate heated to temperatures
membrane resistance. The difference between above 30 C a fine mineral precipitate is formed
curve B and the experimental permeate flux in which causes fouling of UF membranes.
Figure 1 is a measure of the degree of fouling. For the effect of process temperature on
For desalted whey the cause of the fouling may permeate flux, a comparison of Figure 1 and
be due to the high degree of desalting, about Figure 2 shows that the pure water flux in-
95%, because the stability of the whey protein creases by about 3% per degree. The initial
is affected, which may cause a tendency to permeate flux for whey and UF-permeate at 30
aggregate at the membrane surface. C was also about 60% higher than at 10 C, but
For skim milk the differences between due to the strong flux decline at a concentration
experimental and expected (assuming only ratio of about 1.6: 1, the permeate flux at 30 C
concentration polarization) permeate fluxes are decreased to below that at 10 C.
even bigger. The casein in skim milk is respon- For skim milk we see that at 30 C, just as at
sible for considerable fouling, because its high 10 C, the permeate flux was much lower than
concentration and low diffusion coefficient the calculated flux, for only concentration
cause formation of a deposit (gel layer) on the polarization (curve B), so a severe fouling had
membrane surface (10, 27). to be considered. Probably the fouling was
Under these conditions, the osmotic pressure
was the main governing factor for Gouda whey,
UF-permeate, and the lactose solution. In the permeate flux dim 2. h)
case of desalted whey and skim milk a fouling
50 h -. . .-. . -. . . . . pu_re__waterfI_u_x. . . . . . . . . . .
tendency also has to be considered.

Influence of Composition
of Feed at 30 C

A second series of experiments was at 30 C.


In Figure 2 are plotted the experimental
permeat fluxes versus the osmotic pressure of
the bulk liquid. The permeate flux calculated
according to equation [6] of Appendix A, for
only concentration polarization, is presented
also (curve B). For Gouda whey and UF- L L I I L I I

permeate, the permeate fluxes initially resemble 0 .4 .8 1.2 1.6 2.0 2.4 2.8
those calculated. However, at a concentration osmotic pressure (MPa}
ratio of about 1.6:1 (osmotic pressure about 1 Figure 2. Permeate flux versus osmotic pressure of
MPa) a sudden decline in permeate flux was the bulk liquid for RO of skim milk, whey and UF-
found. Obviously considerable fouling of the permeate at 30 C, P = 4 MPa, and v = 2 m/s. Curve B
membrane t o o k place. Since both whey and was calculated according to Appendix A assuming only
concentration polarization (no fouling). • Gouda
UF-permeate show this behavior, it is not likely cheese whey pH = 6.6, × decalcified whey, a UF-
that the whey protein was responsible for this permeate, obtained at 55 C, o UF-permeate, obtained
effect. During the RO of decalcified whey the at 10 C, and o skim milk.

Journal of Dairy Science Vol. 63, No. 2, 1980


208 HIDDINK ET AL.

caused mainly by casein. Since milk is saturated surface may occur. This p h e n o m e n o n agrees
with Ca-phosphate (20), during heating and with that for UF of whey at various pH (9).
concentration a certain precipitation of Ca-
phosphate will take place. The fact that this Influence of Process Conditions
precipitation of Ca-phosphate did not give such
The process conditions such as temperature,
a strong flux decline as in the case of whey and
flow velocity, and pressure have a considerable
UF-permeate may be attributed to the precipi-
effect on permeate flux, and time also can be
tation of Ca-phosphate on casein micelles, and
an important factor. To make the discussion
therefore less precipitation on the membrane.
easier, three resistances to permeation often
In experiments in which UF-permeate,
can be distinguished: the membrane resistance
obtained by UF at 55 C, was concentrated by
Rm, the fouling resistance Rf, and the apparent
RO, the strong flux decline did not occur, the
resistance which can be attributed to concen-
permeate flux being similar to that for decalci-
tration polarization Rp (see Appendix B). The
fied whey (see Figure 2). During UF of whey or
membrane resistance especially is affected by
skim milk at those high temperatures a precipi-
temperature; the higher the temperature, the
tation of Ca-phosphate occurred in the UF
lower the Rm. For the other resistances the
retentate (23). As a result, the UF permeate
matter is more complex as they are particularly
had a lower Ca and phosphate content. In our
dependent on flow velocity and pressure. In the
experiments UF-permeate obtained at 10 C
interpretation of our experiments it is assumed
contained 33 mg Ca and 30 mg P/100 g while
that the effect of time on the permeate flux can
UF-permeate obtained at 55 C had 25 mg Ca
be neglected. This assumption can be justified
and 25 mg P/100 g, which means a reduction of
since each experiment lasted only a few hours.
about 25%. In the latter UF-permeate during
During longer periods the effect of time on the
RO to 30 C apparently no more membrane
permeate flux certainly becomes important.
fouling took place. Roger et al. (22) reported a
similar observation. They found an improvement
Effect of Flow Velocity
in flux of UF-permeate in an enzyme reactor by
removing 15% of the Ca. According to the method in Appendix B, for
Solubility of Ca-phosphate increases at lower
pH (1, 20). At a pH of about 4.9 phosphate
appears to be completely in solution in milk
(20). Some RO experiments were with whey at permeate flux (I/m 2. h)

30 C at various pH. From Figure 3 when . . . . . . p._ur e._w_ater _flEx . . . . . . . . . . . .


50
the whey was acidified to pH 4.6 or even to pH
6.0, the strong flux decline, pH 6.6, did not 40
occur. At pH 4.6 Ca-phosphate is supposed to
be completely soluble so we do not expect
30
Ca-phosphate fouling. At pH 6.0 more Ca-
phosphate is soluble than at pH 6.6. Solubility
20
is improved enough to make possible a fourfold
concentration of whey at 30 C without a strong
10
Ca-phosphate fouling.
The permeate flux (Figure 3) at pH 6.0 was 0
less than the calculated permeate flux (curve l Z 3 4 5
~ i , ~ concentration ratio
B); probably fouling is not avoided completely.
0 14 .8 1.2 1.6 2.0 2.4 2.8
However, the permeate flux at pH 4.6 was osmotic pressure (MPa)
considerably lower than curve B. This effect Figure 3. Permeate flux versus osmotic pressure of
should probably be attributed to the effect of the bulk liquid (concentration ratio) for Re of whey
pH on the behavior of the whey protein (9, 14). at various pH-levels. T = 30 C, P = 4 MPa, v = 2 m/s.
In the neighborhood of the iso-electric point, Curve B was calculated according to Appendix A
assuming only concentration polarization (no foul-
the protein has a low electrical charge. Under ing). • Gouda cheese whey pH = 6.6, DGouda cheese
these conditions aggregation takes place easily whey acidified to pH = 6.0, and o Gouda cheese whey
and accumulation of protein at the membrane acidified to pH = 4.6.

Journal of Dairy Science Vol. 63, No. 2, 1980


REVERSE OSMOSIS OF DAIRY LIQUIDS 209

a range o f e x p e r i m e n t s with skim milk and ratios R f increases. For 2 × c o n c e n t r a t e d skim


whey, R f and Rp were calculated. In Figure 4 milk, Rf at 10 C is m u c h higher than at 30 C; at
t h e e f f e c t o f the flow velocity on R f and Rp is a flow v e l o c i t y o f 2.6 m / s we still have an Rf
given. The main trend is a decrease of Rp and o f a b o u t 1.0 TPa.s/m, which is m o r e t h a n twice
R f with an increase of the f l o w velocity. F o r R m at 10 C. To reach reasonably small Rf at 10
w h e y (2× c o n c e n t r a t e d ) at 10 C, an R f (curve C, m u c h higher flow velocities should be used.
A) of a b o u t .8 TPa.s/m is f o u n d at a f l o w The effect o f f l o w velocity on c o n c e n t r a t i o n
velocity of 1 m/s; this is a b o u t 1.7 × the polarization, expressed as Rp, also is in Figure
m e m b r a n e resistance R m at 10 C. At a flow 4. Curve D represents Rp for w h e y at 10 C and
velocity of 2 m/s, R f has decreased to about skim milk at 30 C (both t w o f o l d c o n c e n t r a t e d ) .
1/5 o f R m . As in Figure 4, a further increase in The Rp is in b o t h cases a b o u t the same since
flow velocity does not diminish R f to any the p e r m e a t e fluxes are at the same level. The
extent. F r o m Figure 4 we see t h a t for skim Rp are considerably less than R f and Rm. It can
milk R f is o f m u c h m o r e i m p o r t a n c e than for be c o n c l u d e d that high flow velocities are n o t
whey. Again d e p e n d e n c e of R f on flow velocity necessary for reducing Rp b u t are particularly
is strong. At 30 C R f for t w o f o l d c o n c e n t r a t e d so for reducing Rf.
skim milk b e c o m e s less t h a n R m at a flow Lim et al. (15) evaluated Rp and R f for RO
velocity o f about 2.4 m/s. An increase in flow of cottage cheese whey, and c a m e to similar
velocity o v e r 2.6 m/s m a y have o n l y a marginal conclusions a b o u t the e f f e c t o f flow rate on Rp
effect. Also presented is R f for a c o n c e n t r a t i o n and Rf. F r o m e x p e r i m e n t s at flow velocities be-
ratio of 1.5 and 3. At higher c o n c e n t r a t i o n tween 15 and 50 cm/s, t h e y c o n c l u d e d that R f
w o u l d b e c o m e negligible at a b o u t 1 m/s. Since
t h e y had to d e t e r m i n e this by e x t r a p o l a t i o n , it
resistance (TPa. s/ml
is understandable that this figure cannot be
accurate.
c\ For skim milk the effect o f the flow v e l o c i t y
3.5 on R f and, thus, on the p e r m e a t e flux is stronger
than for whey. Since Rp is small c o m p a r e d with
3.0 Rf, it may be assumed that R f is the main
governing factor for the p e r m e a t e flux. If so,
the situation is similar to that e n c o u n t e r e d for
2.5 3.0x U F . F r o m theoretical considerations o f UF (19)
1 . 1 .
the p e r m e a t e flux J v or -Rf - (since J v a ,Rf - - ) .Is
20
. B \ \\\ ~+. p r o p o r t i o n a l to the mass transfer coefficient K,

1.5
20
.x? \ \ \ \ \ ~. which is for t u r b u l e n t flow in tubes propor-
tional to Re.8 (see A p p e n d i x A). However, for
1,5x
UF often the effect of flow velocity is m u c h
ON \\ \
1.0 A\ \ \ " greater than when calculated according to this
relationship (17, 19). An evaluation o f our
.5 .._R_mat 10 C ~-x \ "°~.~'~. results with skim milk and of the results of
Rmat30 C ~ \"-,..'~-~''---,- Skudder et al. (27) showed a p r o p o r t i o n a l i t y to
o ~, -,----~---, ......... -,----:==:=-
- ---T-~ Re 1-3. Due to the relatively narrow range o f
.5 1.0 1.5 2.0 2.5 variation in flow velocity this is not accurate,
flow rate qmls) but it emphasizes the similarity b e t w e e n RO of
Figure 4. Various resistances to water permeation skim milk and U F of protein solutions. In both
as a function of the flow velocity. A) Fouling resist- situations a protein layer is f o r m e d on the
ance (Rf) for whey, 2X concentrated, T = 10 C, P =
membrane.
4 MPa; B) Fouling resistance (Rf) for skim milk, 1.5,
2.0, and 3.0X concentrated, T = 30 C, P = 3.5 MPa;
C) Fouling resistance (Rf) for skim milk, 2X concen- Effect of Pressure
trated, T = 10 C, P = 3.5 MPa; and D) Resistance due Particularly for RO of skim milk applied
to concentration polarization (Rp), o for whey 2×
pressure is a critical factor. Skudder et al. (27)
concentrated, T = 10 C, P = 4 MPa, a for skim milk,
2X concentrated, T = 30 C, P = 3.5 MPa (1 TPa.s/m = p e r f o r m e d a range o f e x p e r i m e n t s b e t w e e n 2.4
1012 Pa.s/m). and 4.1 MPa and f o u n d that the p e r m e a t e flux

Journal of Dairy Science Vol. 63, No. 2, 1980


210 HIDDINK ET AL.

passed t h r o u g h a m a x i m u m in this region. decrease a t pressures a b o v e 4 MPa. In Figure 6


W e carried o u t a similar r a n g e o f e x p e r i m e n t s in t h e p u r e w a t e r f l u x also is p r e s e n t e d , a n d t h e
w h i c h t h e applied pressures were varied f r o m p e r m e a t e f l u x w h i c h c o u l d be e x p e c t e d if o n l y
2.5 to 4.5 MPa at 10 C a n d 30 C. F r o m t h e t h e o s m o t i c pressure were t a k e n i n t o a c c o u n t
p e r m e a t e fluxes t h e f o u l i n g resistance R f was (curve B) a n d t h e p e r m e a t e f l u x if t h e e f f e c t o f
c a l c u l a t e d . In Figure 5, R f is p r e s e n t e d as a c o n c e n t r a t i o n p o l a r i z a t i o n were also c o n s i d e r e d
f u n c t i o n o f t h e a p p l i e d pressure. A t 10 C (curve C). T h e p h e n o m e n o n t h a t t h e p e r m e a t e
( c o n c e n t r a t i o n ratio 1.5:1 a n d 2 : 1 ) R f a t 2.5 f l u x is i n d e p e n d e n t o f t h e applied pressure
MPa are c o n s i d e r a b l y l o w e r t h a n R m . F u r t h e r - a b o v e 3 MPa s h o u l d , a c c o r d i n g t o o u r calcula-
m o r e , linear increase o f R f w i t h pressure is tions, b e a t t r i b u t e d t o increasing f o u l i n g
strong. A t 30 C a n d a t a p r e s s u r e o f 2.5 MPa, resistance. A t a p r o c e s s t e m p e r a t u r e o f 10 C
R f a g a i n are small, a n d a b o v e 3 MPa increase o f t h i s e f f e c t is even m o r e s e r i o u s t h a n at 30 C, a
R f w i t h pressure is a l m o s t p r o p o r t i o n a l . Also, c o n s t a n t p e r m e a t e flux over t h e w h o l e p r e s s u r e
R f d e p e n d s o n t h e c o n c e n t r a t i o n ratio (Figure range in this case.
5); t h e h i g h e r t h e c o n c e n t r a t i o n r a t i o t h e h i g h e r T h e r e s p o n s e o f t h e p e r m e a t e f l u x to p r e s s u r e
t h e Rf. This e f f e c t is m o r e p r o n o u n c e d at v a r i a t i o n s can be u n d e r s t o o d as follows. In a
10 C t h a n at 30 C. s t e a d y s t a t e s i t u a t i o n as m u c h s o l u t e ( i n c l u d i n g
In Figure 6 t h e p e r m e a t e flux for skim m i l k p r o t e i n ) is m o v e d b a c k i n t o t h e b u l k liquid -
at a c o n c e n t r a t i o n ratio 1.5:1 is p l o t t e d versus b y d i f f u s i o n a n d t u r b u l e n c e - as is b r o u g h t to
t h e a p p l i e d pressure. T h e a c t u a l p e r m e a t e flux t h e m e m b r a n e surface u n d e r t h e i n f l u e n c e o f
increased to a pressure o f a b o u t 3.0 MPa, a n d p e r m e a t i o n t h r o u g h t h e m e m b r a n e . T h e trans-
for h i g h e r pressures a c o n s t a n t p e r m e a t e p o r t b a c k i n t o t h e b u l k is g o v e r n e d b y t h e mass
flux was r e a c h e d , w i t h a slight t e n d e n c y to t r a n s f e r c o e f f i c i e n t w h i c h again is d e t e r m i n e d
b y t h e flow velocity. If at a c o n s t a n t flow
v e l o c i t y a h i g h e r pressure is applied, d u e to an
initial higher p e r m e a t e flux, t h e t h i c k n e s s o f
foulin( resistance (TPa.s/m) t h e fouling layer increases, w h i l e at t h e same
time a certain compaction of the deposit may
1.4 9/ A t a k e place. T h e s e e f f e c t s will r e s u l t in a h i g h e r
/
/
/
1.2 //
//
/ permeate flux (l/m 2. h)
l.O 2/ .- A
//
/
/
.8 //
/ +,.- B
/ /
9/ /
.6 / / //
Rmat]O
---
C /
/ /+"
/
.4 - // / . ..,..~/ C
Rm at // +// /
tO // / " o x- D
-;0-c , E

2.5 3.0 3.5 4.0 4.5


applied pressure (NtPa) 0 /~ 2.5
~ ~
3.0 a
3.5 i
40 i
45
Figure 5. Fouling resistance as a function of the pressure (/VlPa)
applied pressure for RO of skim milk at a flow velocity Figure 6. Permeate flux as a function of the applied
of 2.6 m/s. A) T = 10 C, concentration ratio 2.0:1; pressure for RO of skim milk (1.5X concentrated) at
B) T = 10 C, concentration ratio 1.5:1; C) T = 30 C, T = 30 C, v = 2.6 m/s. A) pure water flux; B) permeate
concentration ratio 2.0:1; D) T = 30 C, concentra- flux to be expected without concentration polariza-
tion ratio 1.5:1; and E) T = 30 C, concentration tion and fouling; C) permeate flux to be expected
ratio 1.05:1 (1 TPa.s/m = 1012 Pa.s/m). without fouling; and D) actual permeate flux.

Journal of Dairy Science Vol. 63, No. 2, 1980


REVERSE OSMOSIS OF DAIRY LIQUIDS 211

Rf, and this process continues until such a high 2 the results of our e x p e r i m e n t s are given. F o r
R f is built up t h a t equilibrium exists again b o t h layers a fairly strong d e p e n d e n c e on
b e t w e e n the a m o u n t o f solute transported to c o n c e n t r a t i o n ratio, pressure, and flow v e l o c i t y
and f r o m the m e m b r a n e surface. Ultimately was f o u n d , which makes it d o u b t f u l that there
this results in an unchanged p e r m e a t e flux. The was a characteristic difference b e t w e e n t h e t w o
slight decrease in p e r m e a t e flux at higher layers. We think it is m o r e useful to consider
pressures can be explained by a m o r e serious the fouling layer as a whole and to distinguish
c o m p a c t i o n o f the fouling layer. Skudder et al. within the fouling layer a certain gradient in
(27) report that the position o f the m a x i m u m c o n c e n t r a t i o n and firmness. This m o d e l agrees
in p e r m e a t e flux d e p e n d s b o t h on f l o w velocity with that for c o n v e n t i o n a l filtration with a
and t e m p e r a t u r e . The higher the f l o w velocity p o r o s i t y gradient (30). The effects in Table 2
and t e m p e r a t u r e , the lower the R f and the can be u n d e r s t o o d as follows. The higher the
higher the o p t i m u m pressure. f l o w velocity (shear rate), the m o r e will be
In our e x p e r i m e n t s with w h e y an effect of scraped o f f o f the fouling layer and the lower
pressure on p e r m e a t e flux was similar but only the remaining fouling resistance after water
at pressures above 4.5 MPa. Such an effect for rinsing. High pressures will cause a greater
w h e y also was f o u n d by D o n n e l l y et al. (5). c o m p a c t i o n of the fouling layer, which again
The l o w protein c o n t e n t o f w h e y and the causes a firmer and m o r e fouling resistance
high diffusion coefficient for w h e y protein after water rinsing.
c o m p a r e d to casein m a y m e a n that for w h e y at
higher pressures o n l y (i.e., higher p e r m e a t e CONCLUSIONS
fluxes) a considerable fouling resistance is built.
For reverse osmosis of dairy liquids, b o t h
the c o m p o s i t i o n o f the feed and process condi-
Structure of Fouling Layer
tions have an i m p o r t a n t effect on p e r m e a t e
Lim et al. (15) distinguished within the flux. F l u x limiting factors include o s m o t i c
fouling layer ( e x p e r i m e n t s with cottage cheese pressure and fouling tendency. F o r G o u d a
whey), a cake layer which should be a c o m p a c t whey, processed at 30 C, fouling o f the m e m -
layer " a t t a c h e d " to the m e m b r a n e surface, and branes due to Ca-phosphate precipitation is an
a viscous layer b e t w e e n the cake layer and the i m p o r t a n t factor, while for skim milk, desalted
circulating bulk. The characteristic difference w h e y , and acid whey, protein is the i m p o r t a n t
b e t w e e n those t w o layers is that the viscous fouling agent. Fouling by G o u d a w h e y can be
layer is removed by rinsing with water while the limited by decreasing the p H or by decalcifying
cake layer remains a t t a c h e d to the m e m b r a n e . the whey.
In our e x p e r i m e n t s with skim milk there was F u r t h e r m o r e , it is possible, by p r o p e r
also a proteinaceous fouling layer as for Lim et process conditions in which the flow v e l o c i t y
al. (15). This layer was divided into a cake layer plays an i m p o r t a n t role, to decrease the resist-
and a viscous layer (see A p p e n d i x B). In Table ance due to fouling and c o n c e n t r a t i o n polariza-

TABLE 2. Distribution of the fouling resistance over a viscous layer and a cake layer for RO of skim milk.
Rcake = fouling resistance a f t e r w a t e r rinsing. Rviscou s = Rf - Rcake. (1 TPa.s/m) = 1012 Pa.s/m).

Rviscous (TPa.s/m) Rcake (TPa.s/m)


Concentration ratio
1.05 2.0 3.0 1.05 2.0 3.0

P = 2.5 MPa v = 2.6 m/s .01 .01 . . . . 02 .03 ...


P = 3.5 MPa v = 1 . 4 m/s .16 .88 . . . . 63 .66 ...
v = 2.0 m/s .12 .29 .54 .18 .21 .24
v = 2.6 m/s .05 .10 .24 .04 .07 .09
P = 4.5 MPa v = 2.6 m/s .07 .22 . . . . 13 .20 ...

Journal of Dairy Science Vol. 63, No. 2, 1980


212 HIDDINK ET AL.

tion considerably below the resistance of the plants save costs in w h e y processing t h r o u g h
membrane. An important step needed to reach reverse osmosis system. Dairy and Ice Cream Field
160:1160.
a h i g h e r p e r m e a t e f l u x is t h e d e v e l o p m e n t o f
14 Lee, N. D., and R. L. Merson. 1976. Chemical
membranes with higher water permeabilities t r e a t m e n t s o f cottage cheese w h e y to reduce
and resistance to high temperatures. fouling of ultrafiltration m e m b r a n e s . J. Food Sci.
41:778.
15 Lim, T. H., W. L. Dunkley, and R. L. Merson.
1971. Role of protein in reverse osmosis o f cottage
REFERENCES cheese whey. J. Dairy Sci. 54: 306.
1 Brule, G., E. Real del Sol, J. Fauquant, and C. 16 Mehta, B. 1973. Processing o f model compositional
Fiand. 1978. Mineral salts stability in a q u e o u s whey solutions with pressure driven membranes.
phase o f milk: Influence o f heat treatments. J. Ph.D. thesis, The Ohio State University, Columbus.
Dairy Sci. 61:1225. 17 Muller, L. L. 1976. Whey utilization in Australia.
2 de Boer, R., J. N. de Wit, and J. Hiddink. 1977. Australian J. Dairy Technol. 31 : 92.
Processing of whey by m e a n s o f m e m b r a n e s and 18 Peri, C., and W. L. Dunkley. 1971. Reverse osmosis
applications o f whey protein concentrate. J. Soc. of cottage cheese whey. 1. Influence o f composition
Dairy Technol. 30:112. o f the feed. J. Food Sci. 36:25.
3 Dejmek, P. 1975. Permeability o f the concentra- 19 Porter, M. C. 1972. Concentration polarization
tion polarization layer in ultrafiltration o f macro- with m e m b r a n e ultrafiltration. Ind. Eng. Chem.
molecules. Page A2.26 in Separation processes by Prod. Res. Develop. 11 : 234.
m e m b r a n e s , ion exchange and freeze-concentration 20 Pyne, G. T. I 9 6 2 . Reviews o f the progress o f dairy
in food industry. A.P.RA.A., Paris. science. Some aspects o f the physical chemistry of
4 D e l a n e y , R.A.M., and J. K. Donnelly. 1977. the salts o f milk. J. Dairy Res. 29:101.
Applications o f reverse osmosis in the dairy indus- 21 Reid, C. E. 1972. Principles of reverse osmosis.
try. Page 417 in Reverse osmosis and synthetic Page 109 in Industrial processing with membranes.
m e m b r a n e s . S. Sourirajan, ed. Nat. Res. Counc. R. E. Lacey and S. Loeb, ed. Wiley-Interscience,
Can., Ottawa. New York.
5 Donnelly, J.K., A. C. O'Sullivan, and R.A.M. 22 Roger, R., J. L. T h a p o n , J. L. Maubois, and G.
Delaney. 1974. Reverse osmosis-concentration Brule. 1976. Hydrnlyse du lactose c o n t e n u dans
applications. J. Soc. Dairy Technol. 27:128. l'ultrafiltrat de lair ou de lactosdrum en r~acteur
6 Eriksson, P. 1977. Concentration of whey by e n z y m a t i q u e ~ m e m b r a n e . Le Lait 5 5 1 - 5 5 2 : 5 6 .
reverse osmosis-inventory and operating experi- 23 Rose, D., and H. Tessier. 1959. Composition of
ences. Nordeuropaeisk Mejeritidsskrift. 43 : 238. ultrafiltrates from milk heated at 80 to 230 F in
7 Evans, E. W., and F. A. Glover. 1974. Basic prin- relation to heat stability. J. Dairy Sci. 42:969.
ciples of reverse osmosis and ultrafiltration. J. Soc. 24 Schoorl, N. 1929. Suiker titratie. Chem. Weekblad
Dairy Technol. 27:111. 26:130.
8 Fenton May, R. I., C. G. Hill, C. H. A m u n d s o n , M. 25 Short, J. L., and R. K. Doughty. 1976. An evalua-
H. Lopez, and P. D. Auclair. 1972. Concentration tion of m o d u l e s and m e m b r a n e s for the concentra-
and fractionation of skim milk by reverse osmosis tion of cheddar cheese whey by reverse osmosis.
and ultrafiltration. J. Dairy Sci. 55 : 1561. New Zealand J. Dairy Sci. Technol. 11 : 237.
9 Forbes, F. 1972. Considerations in the optimisation 26 Short, J. L., and I. R. Hughes. 1978. The concen-
of ultrafiltration. Chem. Eng.:21. tration of separated milk by reverse osmosis. New
10 Glover, F. A., and B. E. Brooker. 1974. The Zealand J. Dairy Sci. Technol. 13:114.
structure of the deposit formed on the m e m b r a n e 27 Skudder, P. J., F. A. Glover, and M. L. Green.
during the concentration of milk by reverse osmo- 1977. An e x a m i n a t i o n of the factors affecting the
sis. J. Dairy Res. 41:89. reverse osmosis o f milk with special reference to
11 Glover, F. A., P. J. Skudder, P. H. Stothart, and E. deposit formation. J. Dairy Res. 44: 293.
W. Evans. 1978. Reviews of the progress of dairy 28 Smith, B. R., and R. D. Macbean. 1978. Fouling in
science: reverse osmosis and ultrafiltration in dairy- reverse osmosis. Australian J. Dairy Technol.
ing. J. Dairy Res. 45:291. 33:57.
12 Hayes, J. F., J. A. Dunkerley, and L. L. Muller. 29 Sourirajan, S. 1970. Reverse osmosis. Academic
1974. Studies on whey processing by ultrafiltration. Press, New York.
Austr. J. Dairy Technol. 29:132. 30 Tiller, F. M., and J. R. Crump. 1977. Solid liquid
13 J o h n s t o n , A. N. 1977. How European cheese separation: An overview. Chem. Eng. Progr. 73:65.

APPENDIX A somewhat simplified form, by the following


equation:
Calculation of Concentra-
tion Polarization
Jv = (AP- ATrw)/Rm [11
The permeate (water) flux through a reverse
osmosis membrane is u s u a l l y d e s c r i b e d , in a where

Journal of Dairy Science Vol. 63, No. 2, 1980


REVERSE OSMOSIS OF DAIRY LIQUIDS 213

Jv flux through m e m b r a n e ( l / m 2 . h ) or
-- where
(m/s)
AP= hydrostatic pressure difference across
the m e m b r a n e (Pa) Sh = kd/ID ( S h e r w o o d n u m b e r )
ATrw = o s m o t i c pressure difference across the Re = dv//) ( R e y n o l d s n u m b e r )
m e m b r a n e (Pa) Sc = u/ID ( S c h m i d t n u m b e r )
R m = resistance to water p e r m e a t i o n o f the and
m e m b r a n e (Pa.s/m) d = d i a m e t e r of the t u b e (m)
v = f l o w velocity in the t u b e (m/s)
If restricted to high rejection membranes, ID = diffusion coefficient (m2/s)
the o s m o t i c pressure at the p e r m e a t e side m a y v = kinematic viscosity (m2/s)
be neglected; then &Trw equals Zrw, which is the
osmotic pressure at the m e m b r a n e surface at
retentate side. As a result of the rejection of the For the calculation of k, e q u a t i o n (4) can be
solutes by the m e m b r a n e , at the m e m b r a n e sur- written as:
face a local c o n c e n t r a t i o n (and local o s m o t i c
pressure) is found which is higher than in the k = .023 ID '67 v ' 8 / ( d "2 /).47) [5]
bulk of the solution. This c o n c e n t r a t i o n polari-
zation is c o u n t e r a c t e d by diffusion of solutes Equation [31 was evaluated for the liquids
back into the bulk, by t u r b u l e n t eddies, and by referred to in Table 1. For such an evaluation
shearing of the fluid near the m e m b r a n e . At the physical properties o f the liquids m u s t be
equilibrium the transport o f solutes back into k n o w n . The viscosity and o s m o t i c pressure
the bulk occurs as fast as solutes are brought up were d e t e r m i n e d e x p e r i m e n t a l l y , the values o f
to the m e m b r a n e . A t equilibrium the concen- the diffusion coefficient were taken f r o m the
tration o f solutes ( o s m o t i c pressure) at the literature (Table A.1). In our calculations con-
m e m b r a n e surface is given by (7, 19): cerning c o n c e n t r a t i o n polarization, we o n l y
a c c o u n t e d for those c o m p o n e n t s which are
C w / C b = Ztw/~ b = exp (Jv/k) [2] contributing to the o s m o t i c pressure (i.e.,
lactose and salts). Since lactose and salts have
where different diffusion coefficients and, thus,
different k-values, e q u a t i o n [31 was e x t e n d e d
CW =c o n c e n t r a t i o n o f solutes at the m e m - to two c o m p o n e n t s as follows:
brane surface (kg/kg)
C b = c o n c e n t r a t i o n of solutes in the bulk
(kg/kg) Jv = (AP/Rm) - (/Tbl/RM)'exp (Jv/kl)
[6]
Ti'b = o s m o t i c pressure in the bulk (Pa)
k= mass transfer c o e f f i c i e n t (m/s) - ( r r b 2 / R m ) - e x p (Jv/k2)

It is assumed that the o s m o t i c pressure is directly It appeared that an evaluation o f e q u a t i o n [3]


p r o p o r t i o n a l to the solute concentration. A resulted in a p e r m e a t e flux versus o s m o t i c
c o m b i n a t i o n of equations [1] and [2] results pressure of the bulk liquid which nearly coin-
in: cides for w h e y , U F - p e r m e a t e , and lactose, see
curve B in Figure 1. The variations were less
Jv = (AP - 7rw)/Rm than 1 liter/m 2.h. This can be u n d e r s t o o d f r o m
[3] the fact that the mass transfer c o e f f i c i e n t k
= (& P / R m ) - ( z r b / R m ) exp (Jv/k) for these liquids is about equal which again is
caused by t h e fact that the process c o n d i t i o n s
If k is k n o w n , the p e r m e a t e flux J v can be calcu- are the same and the viscosity, density and diffu-
lated. F o r a calculation o f k, for t u r b u l e n t flow sion coefficients do not differ much. The calcu-
in tubes, generally the following dimensionless lated p e r m e a t e flux for skim milk tends to be
correlation is used (19): s o m e w h a t lower (.5 to 2.0 l i t e r / m 2 . h ) than
curve B in Figure 1. This can be u n d e r s t o o d
Sh = .023 Re.8 Sc.33 [41 f r o m the higher viscosity o f skim milk.

Journal of Dairy Science Vol. 63, No. 2, 1980


214 HIDD1NK ET AL.

TABLE A . I . Solute-water diffusion coefficients for some components of milk at 20 C.

Diffusion
coefficient (m2/s) Reference

Lactose 5% 3.8 × 10-10 (4)


Ash(KCI.74%) 18.4 X 10 -1° (29)
Whey protein (/3-1actoglobulin 1%) .64 X 10 -10 (4)
Casein 3% .19 × 10 -1° (4)

APPENDIX B Jv = (AP - rrw)/(Rm + Rf) [71


Evaluation of Resistance due to
Fouling and
Concentration Polari- Next the c o n c e n t r a t i o n polarization resistance
zation from the Experiments Rp can be calculated f r o m the following equa-
tion:
Strictly speaking, c o n c e n t r a t i o n polarization
reduces the driving force. However, the effect
Jv = (AP - "trb)/(Rm + Rf + Rp)
of c o n c e n t r a t i o n polarization o f t e n is translated
into an additional resistance to p e r m e a t i o n o f
To divide the fouling resistance (Rf) in a re-
the m e m b r a n e , when the original driving force
sistance of a cake layer (Rcake) and the
is considered. However, a fouling layer on the
resistance of a viscous layer (Rviscou s) the
m e m b r a n e m a y cause a real additional hydraulic
following procedure was followed. The bulk
resistance acting in series with the m e m b r a n e
liquid was r e m o v e d by water; then the pure
resistance (28).
water flux was measured under the same
The m e m b r a n e resistance R m can be deter-
c o n d i t i o n as previously used for t h e liquid.
mined f r o m the pure water flux for the clean
F r o m this pure water flux Jw, Rcake was
membrane.
calculated according:
F r o m the e x p e r i m e n t a l p e r m e a t e flux for
the liquid concerned, after an evaluation of the
mass transfer coefficient k, the c o n c e n t r a t i o n J w = A P / ( R m + Rcake ) [8]
polarization modules Cw/Cb = rrw/Zrb can be
calculated f r o m A p p e n d i x A, e q u a t i o n [21. N e x t Rviscuu s can be calculated f r o m :
If k n o w n nw the value of the fouling resist-
ance R f can be calculated since Rviscou s = R f - Rcake [9]

Journal of Dairy Science Vol. 63, No. 2, 1980

You might also like