Comparing DOE & DCT

balanceWhen selecting a subset of parameters, Design of Experiments (DOE) relies on the experience of the individuals involved and that introduces a number of significant problems.  The first is that, because it is based entirely on the judgment of individuals, it allows the influence of bias, be it conscious or not.  The second problem is that DOE, when applied to a new or complex system, is very inefficient.  Part of this inefficiency stems from the step-by-step or piecemeal approach that DOE requires which is of necessity time consuming.  The final problem is that DOE, because it relies on a trial and error method of assessing progress, is grossly inefficient, particularly when applied to complex systems.

With DOE, the operator applies his or her experience and judgment to a process, breaking it down into a series of parameters in order to simplify the overall system for testing but by breaking the overall process down into individual sub-processes DOE can generate misleading or inaccurate results.    Additionally, because it relies entirely on traditional statistical principles and methods, DOE requires many experiments to be conducted sequentially in order to optimize a complex system.  This requirement not only increases the risk of bias, it is also dependent on the assumption that earlier changes made against the background of a specific environment remain valid once that environment has been changed by later experiments.

Comparing Design of Experiments (DOE) and Directional Control Technology (DCT)

DOE DCT
A detailed analysis of parameter characteristics is required to create a subset of parameters that will be used in subsequent experiments. No detailed analysis of parameter characteristics is required.
Expert knowledge required to choose a subset of parameters. Minimal expert knowledge required when choosing parameters and no subset is required.
Fragmentary approach may unintentionally exclude important or include unimportant parameters. Comprehensive approach guarantees that all parameters are tested in initial diagnostic experiments. Parameters are then included or excluded based on their contribution to the objective.
For a 60 parameter system requires the system to be broken down into much smaller segments. Ultimately hundreds or possibly thousands of experiments would be required. Typically requires no more than 10 experiments even for a large 60+ parameter system.
Relies on traditional statistical methods which may introduce bias. A systems engineering technology which offers a more reliable and comprehensive approach.
Can only optimize conditions or substances separately. Can optimize conditions and substances simultaneously
A time-consuming and repetitive process. A rapid efficient process.

* very complex systems may require more than 10 experiments