Modelling Approaches for Bio-Manufacturing Operations - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 97

the size of the experimental study. For a given unit operation, there are normally a
set of operating parameters which are conventionally known to be potentially
significant, and this list can be supplemented using information gained during
process development, e.g. by using Ishikawa diagrams or failure modes and effects
analysis (FMEA). Subsequently, one would conduct a series of factorial designs, at
each stage with increasing density of test points and/or fewer variables, as one
seeks to characterise the process design space with respect to its controlling
parameters. From a quality-by-design perspective, such an approach would be
carried out using qualified small-scale models to generate the empirical data; For
example in one study, Looby et al. [ 27 ] employed a combination of FMEA, small-
scale experiments, analysis of variance (ANOVA) modelling and Monte Carlo
analysis to define a process design space to identify operating conditions that
minimised the risk of manufacturing out-of-specification material. Thus design of
experiments (DOE) can help to achieve thorough process characterisation and gain
valuable understanding of which parameters are important to control, along with
their acceptable ranges. This information can be particularly useful in uncovering
interactions between steps and ultimately can help to achieve the most robust
manufacturing strategy that tolerates process variations without compromising on
product quality.
3.4 Business-Process Models
The types of models discussed above have focussed predominantly upon creating
mathematical relationships to connect process parameters with technical manu-
facturing outcomes such as recovery, clarification, purity and quality. Another
class of models considers financial aspects instead to evaluate process economics
in terms of both capital and running expenditure. Ideally, such models must
examine not only the main manufacturing tasks but also ancillary activities such as
equipment, intermediate or buffer preparation, cleaning-in-place, steaming-
in-place and validation. Such facility- or process-related issues must be looked at
in context with strategic or corporate priorities such as clinical supply schedules.
Although business and process concerns have often been seen and modelled as
separate aspects, in reality they must be connected quantitatively to deliver the
most efficient and cohesive process development programme. Thus recently, there
has been an attempt to combine these model types together to perform process
flowsheet evaluations from both perspectives. Such models evaluate both direct
and indirect process expenditure by assigning costs to objects such as resources
and then accumulating a cost within the simulation environment every time a
resource item is either used or purchased (e.g. for consumable items such as resins
or membranes). Other costs may be calculated as a fraction of capital investment
or running costs (e.g. for estimation of overheads, depreciation, maintenance or
taxes). Visualising these costs on either a unit operation basis (i.e. for each process
step) or on a cost category basis (e.g. buffers, labour, ancillary items, overheads,