IDM | Technology/IT
or ultrafiltration (UF) permeates from skim milk and sweet or acid whey
are potential media for DF. The question is what the impact of the different
properties and compositions of such media is on the deposited
layer structure and, thus, on the filtration performance.
Effect of preconcentration on time and
DF medium demand
Equation (1) allows to calculate the required time to achieve a certain purity
described as residual content of the permeating components (Ct,DF/CC0,DF)
during a continuous diafiltration process. It is obvious that the time (t) decreases
when either the initial volume V0,DF or the desired degree of purity
decrease, or on the other hand the permeation (p) or flux (J) increase.
(1)
The permeation p of the target component expected to be transmitted
into the filtrate or permeate (i.e. the whey proteins in this case)
is defined by equation (2):
(2)
The conclusion from equation (1) would theoretically be to concentrate
the starting material as much as possible to decrease V0,DF prior to the
diafiltration process. However, it is known that the flux decreases with
increasing concentration factor (Kersten 2001), which means there is
an optimum between starting volume reduction and flux performance.
A first step was to determine to what extent a concentration of
the milk prior to the DF process minimizes the time for the DFprocess.
For this purpose, pasteurized skim milk was concentrated
up to a factor (CF) of 6 and the flux as well as the permeation
of β-lactoglobulin (β-Lg), as the major whey protein fraction, were
measured. In addition to a continuous flux decline it can be seen
from Fig 2, that the permeation stays constant after an initial decline
across the duration of the pre-concentration step. This means
that the permeability of the deposit layer for β-Lg is not affected
by the increasing protein concentration.
Therefore, the time required for the DF step in a subsequent DF
process can be calculated as a function of the concentration factor
(flux/retentate volume) (Fig. 3).
Fig. 2 Flux and permeation as a function of the concentration
factor.
18 · 8 2017 | international-dairy.com
Fig. 3 Time demand for one DF step as a function of the preconcentration
of the starting material prior to diafiltration.
It can be seen from Fig. 3 that the time required for a DF step
decreases by 20% to a CF of approx. 2.5 and then rises sharply. This
means that the observed flux decline (Fig. 2) is over-compensated
disproportionately by the volume reduction. From CF 2.5, the ratio
reverses and the flux decreases disproportionately to the volume
reduction. This increases the time required for one DF step.
The conclusion is that the total DF time can be reduced to a minimum
when the milk is concentrated to a factor of 2-3 before starting
the DF mode. If the use of a small amount of DF medium is crucial, it
may be useful to concentrate higher and to accept a longer DF time.
Characterization of process side
streams as potential DF media
In order to determine differences in the composition of available DF media,
more than 100 industrial samples from over 30 different processes were
analyzed. Tab. 1 shows some relevant properties of the investigated media.
It is obvious from these data that the different media strongly differ
in their pH value, ionic composition and lactose content. Moreover, it is
known that the casein micelle is strongly affected by different milieu
conditions Broyard and Gaucheron 2015. The hypothesis drawn from
these results was that the use of different media should affect the filtration
performance during the milk protein fractionation. To investigate
this in detail, filtration experiments were carried out using deionized
water, local hard tap water (18 dGH) and UF permeate obtained
from skim millk as DF medium. The reason for choosing these media
was that deionized water is similar to the composition of reverse osmosis
permeate and vapor condensates, tap water is considered as a
reference and UF permeate reflects the natural milieu of milk. Fig. 4
shows the remaining concentration of β-Lg in the MF retentate as a
function of time using these different DF media. Circles are indicating
the times at which the same DF step was accomplished.
It can be seen that after five DF steps a comparable depletion
of whey proteins to approximately 10% of the initial content was
achieved with all DF media. However, there are considerable differences
in the required time to achieve the same depletion effect. The
total process time using UF permeate was increased by 29% com