Implementation of Design of Experiments projects in industry

Abstract

Although design of experiments (DoE) is a common feature of statistics and quality literature, it is insufficiently used in industry. Surveys and numerous articles alike have verified that a gap exists between theoretical development and its effective application in industry. Therefore, we have developed a complete, friendly and easy-to use methodology, from an engineering point of view, to ease the implementation of DoE in companies. Our approach presents a framework for the experimentation process that follows the traditional DMAIC steps of 'Six Sigma' as a generic problem solving methodology. We aim to present a complete overview of the experimentation process useful for those engineers, statisticians and scientists who are not experts in DoE. IMPLEMENTATION OF DoE PROJECTS IN INDUSTRY 479 methodology. Furthermore, 76% of the respondents believed that a methodology is truly needed and that the absence of a clear methodology was one of the main barriers in the application of DoE. Similar results were obtained in other European Regions [6]. In Tay and Butler's [7] research of the existing methodologies for experimental design, they reached the conclusion that two types of DoE methodologies are in use today, namely: • Classical techniques: Developed in the U.K. and U.S. in the 1920s, with the introduction of full factorial designs and fractional designs some years later. In the 1950s, Box-Wilson developed what was later called RSM. Finally, optimal design has been under development since the 1960s. Since then, many have contributed to the development and expansion of this technique. The research of Box et al. [8], Myers and Montgomery [9] and Atkinson and Donev [10] is considered to be fundamental in their fields, respectively. • Taguchi methods: Developed by Genechi Taguchi in Japan in the 1950s, promoting statistical techniques for quality from an engineering perspective. The original work of Taguchi is documented in his own two-volume book [11]. Other books that present the method include Taguchi et al. [12] and Roy [13].After a thorough analysis of their characteristics, Tay and Butler [7] concluded that there was a need to shape an experimental design methodology to develop the information necessary for good planning, design and analysis. Since the majority of those who apply DoE are not statisticians, it is necessary to develop simple methodologies to solve the problems faced in the industry. The Tanco et al. [5] survey echoes this need, showing that 76% of the respondents believed that a methodology is truly needed.Starting from these premises, we developed a complete methodology to make the implementation of DoE in companies easier. We aim to present to industries a simple methodology from an engineering point of view, without ignoring either the statistical or the overall complexity of this task. This methodology will be useful as a complete overview for those who are not experts in DoE and need to carry out an experimentation process. Moreover, it will also be useful for trainers and consultants when it...

A Fast-and-Robust Profiler for Improving Polymerase Chain Reaction Diagnostics

“…To reach to robust decisions, equally important is the analysis procedure for the OA-collected data in the DOE framework [28] . Implementation issues in DOE studies as well as their diverse applications in the fields of industry and engineering have been comprehensively researched [29] , [30] . For applications in biotechnology in particular, there is also an extensive account about the strengths and the weaknesses of Taguchi-related DOE methods [31] .…”

Optimization of Graphite Morphology in Mottled Nihard Cast Irons Inoculated with Feb and Manufactured by Centrifugal Casting

“…Full factorial designs of experiments require a large number of experiments, which grows exponentially depending on the number of factors studied. When there are k factors in a full factorial design of experiments (DOE), the number of tests is 2 k , where 2 is the number of levels applied to each of the factors [20,21]. Fractional factorial DOEs allow a large number of factors to be studied by means of a much smaller number of experiments, assuming the loss of information of possible interactions between factors, which are not usually very significant in practice.…”

Gaussian Process Model – An Exploratory Study in the Response Surface Methodology

“…Each one of these issues has been discussed in the literature, namely, in the classical textbooks by Box et al and Myers et al , where RSM is comprehensively exposed. Authors who discussed non‐statistical issues and presented guidelines for designing and conducting experiments are Simpson et al , Freeman et al , Tanco et al , and Costa et al …”