Once the system requirements, CTQ’s, are derived, the product design can start. This phase results in a high level design, the ‘product architecture’, for the selected concept.
Concept generation and selection
The first task in this phase is the generation of a range of alternative design concepts. Unfortunately, it still frequently occurs that only one concept is generated without considering any alternatives. This results in missed opportunities for better concepts.Next the winning concept is selected based on estimated system performance, costs, reliability, manufacturability and risks. The concept selection is reviewed with all stakeholders, including suppliers, customers and users. This leads to the winning design concept which may be a mixture of several other concepts. The winning concept is turned into a high level design consisting of drawings and a functional prototype. A frequently used tool for concept selection is the Pugh matrix.Paying much attention to generate many alternatives and to review them with a broad audience reduces the probability that the development team, or even worse a competitor, will discover a superior concept much later in the development process.
Once there is a concept design, we need the first predictions to what extent the new design will meet the system requirements. Typical for DfSS is that not only the nominal design is considered, but that also production spreads are taken into account. This results in early predictions of:
- the % of products that will meet the specifications at mass-production and
- the capability indices Cpk and Cp at mass-production
In the traditional engineering approaches, only nominal designs are studied in the early stages of product design. The effect of spreads are postponed to later development phases. However, it turns out that these spreads are the biggest causes for quality problems and hence customer dissatisfaction. Moreover, at later development stages, it turns out to be difficult or costly to change the designs. So, taking the spreads into account in the early phases will prevent costly redesigns or technical and quality problems later on. An example is given in the cake-mix-case.
In the traditional engineering approaches the reliability of the chosen design is often neglected or given low priority in the Design phase. However, as with the capability prediction, this is the stage to assess and prepare for product reliability to avoid costly changes in later development stages. It pays off to collect reliability data of related products and components and build models for life-time or degradation in a similar fashion as for the capability.
Besides the capability prediction a risk analysis for the proposed solution is made. Failure Mode and Effect Analysis (FMEA) is used to assess any risk to the Customer that may have been missed during the project work or needs to be considered during the implementation of the solution. Also the remaining non-product related risks of the project need to be mapped, i.e. people, communication and implementation related issues.
The concept selection results in a breakdown of the system into ‘building blocks’. Building blocks are functional components that can be independently developed and tested. For example, for a coffee machine, the boiler, the pump, the water tank, the brewing chamber, and the coffee pad are building blocks. For the capability prediction the relations between all the building-block output-parameters and the end-product parameters, CTQ’s, are quantified. This quantification is in the form of a mathematical function y=f(x) where the x’s represent all building-block output-parameters and y represents all end-product parameters. Such a function is called a transfer function. Using realistic data for the building-block-parameters, the transfer functions can be used to predict the spread and hence the capability of the design.Related to this is the problem of making a tolerance break down: the specifications of the end product has to be divided over the building blocks. Although it is a standard challenge during product development, it is frequently done incorrectly or insufficiently quantified.
Transfer functions in case of reliability model the performance of building blocks and potentially the total system over time and/or as a function of use conditions.Building transfer functions y=f(x) that quantify the relations between the building blocks and the system, predicting the performance of the end product during mass production and setting balanced tolerances are core competences of CQM.Our Tolerance Designer software (/tolerance-designer) and Compact software (/compact) are tools that facilitate during the design phase the tolerance break down and building the transfer functions respectively. A template for a Pugh matrix can be found at /free-templates.