Neural computing in pharmaceutical products and process development - Computer-Aided Applications in Pharmaceutical Technology - page 161

Benefi ts of the SOMs usage with industrial data were presented
for a set of process measurements in an industrial gas treatment
plant. The practical goal was to identify signifi cant operational
modes and most sensitive process variables before developing an
alternative control strategy. The results confi rmed that the SOM-
based approach is capable of providing valuable information and
offers possibilities for direct application to other process monitoring
tasks (Corona et al., 2010).

Example 4

SOMs were applied for analysis of fl uidized bed granulation
processes (Petrovi ´ et al., 2011) and the study demonstrated the
potential of SOM application in optimization of certain properties
of formulation/product being developed.
SOMs are the basis for identifi cation of a correlation between
the studied properties. For example, if fl owability of the obtained
granules is of interest, two circles marked on the V9 maplet of
Figure 5.20 represent granules having excellent fl owability (Hausner
ratio equals 1.1). At the same time, these two nodes are marked on
the rest of the maplets, in order to identify other properties of the
granules, as well as processing conditions required to produce
granules of excellent fl owability. It can be seen that fl owable granules
are relatively large (200 μm in size) and porous (70%), containing up
to 2% of residual moisture. If we look at the processing parameters,
the following conditions should be set up in order to obtain fl owable
granules: product temperature should not exceed 40 °C, whereas
outlet air temperature should be lower than 35 °C. However,
approximately 0.3 kg of liquid-binder should be consumed with a
relatively high spray rate of liquid-binder (20 g/min). Spray rate
should not exceed 1 bar and drying time does not have any infl uence
on granule fl owability. These fi ndings justify the appropriateness of
the SOM technique in preliminary characterization and optimization
of process and product parameters.

￿

￿ ￿