Evolutionary Operation (EVOP)
By Joaquin G. Zoque, P.Eng.
In recent years, some awareness has been gained about the importance and potential of design of experiments (DOE). Especially since the mid '80s when the Taguchi methods became popularized by the Ford...
In recent years, some awareness has been gained about the importance and potential of design of experiments (DOE). Especially since the mid ’80s when the Taguchi methods became popularized by the Ford Motor Co., this subject has been practiced almost exclusively by statisticians. The success of these techniques relies on in-depth knowledge and command of both technology in the field of application, e.g. extrusion, and DOE statistics. I have seen a consultant failing to produce meaningful results from an extensive factorial experiment, which was unable to include the controlling factors related to the problem the study was supposed to resolve. Several brainstorming sessions with technical staff and manufacturing did not help because they did not have enough command of the technology involved, much less the consultant that conducted the experiment.
Design of experiments is a tool better suited for the pilot plant or development lab in process and product R&D. There, dramatic changes in the result of trials are desired to learn about a system, but such changes can be costly and disruptive if applied in a manufacturing process.
A technique known as Evolutionary Operation (EVOP) is better suited to the manufacturing environment. It allows improvement and optimization of a process by small and gradual changes in parameters, and shows trends for improvement without causing more waste or off-spec product. The trade-off is that it takes a far greater number of trials conducted on real production time. Trials are repeated in cycles until statistical justification is obtained to move to a new phase, a new central operating point, and one phase can take several weeks. The experiment is full factorial two-level, normally run for two factors with a central point and thus allows investigation of curvilinear effects. EVOP has its own special method of analysis and can be very effective in the manufacturing plant.
Joaquin G. Zoque, P.Eng., Oakville, Ont., consults on polymer extrusion, diagnosing opportunities and implementing solutions to increase profitability and productivity.
Tel: 905-847-7080; E-mail: firstname.lastname@example.org