The formation of fouling deposits in heat treatment equipment such as pasteurisers and
evaporators used in milk processing plants reduces heat transfer and is a source of
economic loss as a result of more frequent shut-down of the equipment for cleaning.
Measurement of the kinetics and mechanisms of such fouling will enable an improved
understanding of the design and operation of heat treatment equipment and allow
improved control of fouling.
The present study was carried out to investigate the fouling behaviour of heated whole
milk on to non-heated stainless steel surfaces and to determine the effects of mass flow
rate and flow disturbance on the fouling of these surfaces.
A pilot-scale fouling rig was designed and built to mimic the heat treatment methods
and conditions that are currently used in large-scale processing plants. Whole milk was
first heated from about 6 to 75 oc in a plate heat exchanger and then heat treated to
9 5 o C via a direct steam injection (DSI) heater. Custom built tubular and sudden
expansion fouling test sections of different sizes were placed at different locations
downstream of the D SI heater. These test sections were easily disassembled to access
the fouling layers. Milk flow in these test sections was in the range of 80-120 kg/h and
Reynolds numbers ranged between 2500 and 3800.
A system to monitor fouling at different locations on the internal surfaces of the fouling
test sections was developed. This included the use of a calibrated sensor to measure the
local heat flux and temperature at the outer surfaces of the test section. The fouling rate
was calculated and expressed as the rate of decrease of the internal heat transfer
coefficient during the deposit growth, normalised using the internal heat transfer
coefficient determined at the start of a run. A plot of the fouling rate versus run time
exhibited a delay period during which the normalised internal heat transfer coefficient
was either constant (i.e. the ratio = 1 ) or slightly increased (i. e. the ratio> 1 ) fol lowed
by a decrease period during which it gradually decreased (i. e. the ratio < 1 ). The
decrease period started when the deposit thickness reached about 0. 5 mm, at which stage
the sensor began to respond to the changes in the heat flux and surface temperature.
To investigate the efiects of local flow disturbance on the fouling behaviour of whole
milk heated to 95 o c, three tubular sudden expansion geometries, ratios of upstream-to downstream
diameter of 0.2, 0 . 43 and 0. 74, were used. Fouling in the area beyond the
step of the expansion coincided with the recirculation flow zone characterised by the
rapid change of the fluid shear rates, as shown by the results of the computational fluid
dynamics (CFD) modelling. The local fouling rate in this zone increased when the fluid
velocity was increased and when the ratio of upstream-to-downstream diameter was
decreased (i. e. the step height was increased). These results indicate that flow
disturbance, which in turn causes the fluid velocity to change its direction and/or
magnitude, can induce fouling.
Fully denatured and aggregated P-lactoglobulin or the casein proteins were found to
play a minor role in the fouling of whole milk heated to 95 a C in the fouling rig. It is
therefore hypothesised that, during heating, the native P-Iactoglobulin unfolds to form
an 'active species' and this 'species' is responsible for the fouling of non-heated
surfaces. The concentration of the 'active species' formed during heating is directly
related to the concentration of residual native P-lactoglobulin. The local fouling rate
in various areas downstream of the DSI heater, which was found to relate to the local
concentration of residual native P-lactoglobulin confirms this hypothesis.
Relationships between local and average fouling rates to account for the effects of flow
geometry were determined.
Fouling by heated milk fluids in the downstream areas of the DSI heater in industrial scale
milk evaporators was monitored with heat flux sensors. The fouling behaviour
matched very well the pattern obtained in the fouling rig although the milk was heated
to a higher temperature, i.e. 105 oC, at a higher flow rate, i.e. Reynolds number> 1 05,
and the processing run time of up to 24 hrs was considerably longer than in the pilot-plant
The rate and extent of fouling in large-scale plant were found to vary with the types of
milk product, such as skim milk, whole milk, modified skim milk and modified
whole milk, and with treatment temperatures. Local equipment geometry in the
section downstream of the DSI, such as flow disturbed by an orifice, was also found to
have a strong effect on local fouling.
Recommendations for improving the design and operation of the D SI heater to reduce
fouling of unheated surfaces by whole milk heated 95 o C are suggested