Working on process development with combinatorial methods adds new dimensions to the feedback loop, which consists of (1) the design of the experiment, (2) the experiment itself, and (3) the analysis and the design of the experiment that follows. Traditionally, one cycle is passed through and only one parameter is varied at a time until the process is optimized (Fig. 30.8) . In contrast, the parallel approach enables a large number of parameter sets to be tested simultaneously by changing the screening strategy to fewer, but more diverse and data rich, feedback loops. The extreme example would be an open loop, where all parameter sets can be screened at once and only the best are selected.
With the full range of options of parallel process development available, one has to think about efficiency. ''What is the most advisable ratio between the number of experiments and the number of loops?'' Efficiency can roughly be represented by the product of the frequency (loops), the diversity (number of experiments), and the learning factor (see the formula in Fig. 30.8).
The learning factor includes, for example, the amount of information gained by a single experiment. The amount of information from a single experiment is inversely related to its scale, the complexity of the screening set-up, and its degree of
Was this article helpful?