Robust Design Methodology - a treatise on the gap between principles and tools

Licentiate thesis, 2006

Today many companies are struggling with issues related to unwanted variation. Even though Robust Design Methodology is a suitable means of counteracting such issues there are rather few industrial applications documented. Previous research indicates that this phenomenon might be due to a hitherto existing focus on appropriate statistical tools for achieving robustness in the sense of insensitivity to noise factors. This focus may have led to a negligence of the overarching framework of RDM with its underlying principles: awareness of variation, insensitivity to noise factors and continuous applicability meaning that efforts are applicable in all stages of the product design process. Instead of focusing on these essential principles, there has been a tendency of being taken up with the statistical details of tools associated with RDM, such as e.g. Design of Experiments. In fact, tools associated with RDM are spread in industry; anyhow there is limited use of the methodology which might indicate a gap between RDM and its tools. What might be missing is a link bridging this gap: a link which ensures that the underlying principles or RDM are reflected in the application of appropriate tools. The overall purpose of this thesis is to clarify and elaborate on the gap between the principles underlying RDM and associated tools available. An investigation of a Design of Experiments application in a company working with RDM shows that the tool is used in a suboptimal way meaning that the underlying principles of RDM are not reflected sufficiently in the way the experiments are designed. Four practices constituting a link that can support the realization of those principles are addressed: (1) appreciate Taguchi’s concept of the quadratic loss function, (2) develop P-diagrams and categorize noise factors, (3) systematically look for and utilize nonlinearities between control factors and response by use of e.g. transfer functions, (4) use creativity to achieve concepts insensitive to noise factors. The thesis furthermore clarifies when and how efforts in line with RDM can be applied in the context of a common product design process (PDP). Efforts in line with RDM should be taken up early in the PDP since this is where the foundation of design is laid. Tools are necessary aids in RDM but cannot automatically generate robust designs. Important drivers for implementing and working with RDM are an awareness of variation and thinking in terms of robustness.