Step 4: Reflect on the Results and the Process
The team may require some iteration to agree on the targets. Reflection after each iteration helps to ensure that the results are consistent with the goals of the project. Questions to consider include:
• Are members of the team "gaming"? For example, is the key marketing representative insisting that an aggressive value is required for a particular metric in the hopes that by setting a highgoal , the team will actually achieve more than if his or her true, and more lenient, beliefs were expressed?
• Should the team consider offering multiple products or at least multiple options for the product in order to best match the particular needs of more than one market segment, or will one "ave rage" product suffice?
• Are any specifications missing? Do the specifications reflect the characteristics that will dictate commercial success?
Once the targets have been set, the team can proceed to generate solution concepts.
. The target specifications then can be used to help the team select a concept and will help the team know when a concept is commercially viable. (See Chapter 7, Concept Generation, and Chapter 8, Concept Selection .)
Setting the Final Specifications
As the team finalizes the choice of a concept and prepares for subsequent design and development, the specifications are revisited . Specifications that originally were only targets expressed as broad ranges of values are now refined and made more precise.
Finalizing the specifications is difficult because of trade-offs- inverse relationships between two specifications that are inherent in the selected product concept. Trade-offs frequently occur between different technical performance metrics and almost always occur between technical performance metrics and cost. For example, one trade-off is between brake mounting stiffiless and mass of the fork. Because of the basic mechanics of the fork structure, these specifications are inversely related, assuming other factors are held constant. Another trade-off is between cost and mass. For a given concept, the team may be able to reduce the mass of the fork by making some parts out of titanium instead of steel. Unfortunately, decreasing the mass in this way will most likely increase the
manufacturing cost of the product. The difficult part of refining the specifications is choosing how such trade-offs will be resolved.Here, we propose a five-step process:
I. Develop technical models of the product.
2. Develop a cost model of the product.
3. Refine the specifications, making trade-offs where necessary.
4. Flow down the specifications as appropriate.
5. Reflect on the results and the process.
Step 1: Develop Technical Models of the Product
A technical model of the product is a tool for predicting the values of the metrics for a particular set of design decisions. We intend the term models to refer to both ana lytical and physical approximations of the product. (See Chapter 14, Prototyping, for further dis cussion of such models.)
At this point, the team had chosen an oil-damped coil spring concept for the suspen sion fork. The design decisions facing the team included details such as the materials for the structural components, the orifice diameter and oil viscosity for the damper, and the spring constant. Three models linking such design decisions to the performance metrics are shown in conceptual form in Exhibit 6-9. Such models can be used to predict the product's performance along a number of dimensions. The inputs to these models are the independent design variables associated with the product concept, such as oil viscosity, orifice diameter, spring constant, and geometry. The outputs of the model are the values of the metrics, such as attenuation, stiffness, and fatigue life.
Ideally, the team will be able to accurately model the product analytically, perhaps by implementing the model equations in a spreadsheet or computer simulation. Such a model allows the team to predict rapidly what type of performance can be expected from a particular choice of design variables, without costly physical experimentation. In most cases, such analytical models will be available for only a small subset of the metrics. For example, the team was able to model attenuation analytically, based on the engineers' knowledge of dynamic systems.
Several independent models, each corresponding to a subset of the metrics, may be more manageable than one large integrated model. For example, the team developed a separate analytical model for the brake mounting stiffness that was completely independent of the dynamic model used to predict vibration attenuation. In some cases, no analytical models will be available at all. For example, the team was not able to model analytically the fatigue performance of the suspension, so physical models were built and tested. It is generally necessary to actually build a variety of different physical mock-ups or prototypes in order to explore the implications of several combinations of design variables. To reduce the number of models that must be constructed, it is useful to employ design-of-experiments (DOE) techniques, which can minimize the number of experiments required to explore the design space. (See Chapter 15, Robust Design, for a summary of DOE methods.)
Armed with these technical models, the team can predict whether any particular set
of specifications (such as the ideal target values) is technically feasible by exploring different combinations of design variables. This type of modeling and analysis prevents the team from setting a combination of specifications that cannot be achieved using the available latitude in the product concept.