Home Random Page


CATEGORIES:

BiologyChemistryConstructionCultureEcologyEconomyElectronicsFinanceGeographyHistoryInformaticsLawMathematicsMechanicsMedicineOtherPedagogyPhilosophyPhysicsPolicyPsychologySociologySportTourism






General Robust Design Model

I. CHAPTER ROAD MAP

In this chapter, we first develop the underlying objectives and modeling theory for robust design (Figure 19.1).

 

II. QUALITY DESIGN THEORY

 

 

 

General Robust Design Model

http://www.youtube.com/watch?v=Qz_XuLfd9eA

https://www.youtube.com/watch?v=mPUzQolrcD4

Consider a product as a black box input-output model, the simplest form of the functional model of Chapter 5. The product uses materials, information, and energy to produce output. A designer can apply a performance specification to the difference between the inputs and outputs, the performance variables. These are the outputs to be ensured by the designer.

To do this, the product designer must make configuration and parametric choices to specify the product configuration, a complete dimensional, material, and manufacturing description. These choices are the design variables. It takes a skilled engineer to determine what makes effective design variables-what variables can be specified that impact the variation of the performance variables?

The variation in the performance variables is caused by something. Materials into the manufacturing process, deviations in the operations of production, and differences in the working or user environment all cause the performance of a product to not be on target. These causes of variation are the noise variables. The reason selecting a different design configuration can improve robustness is that the sensitivity of the performance variables changes.

This idea is depicted in Figure 19.2. Here, a performance variable y is graphed against a design variable x that has super-imposed on it an additional uniform noise variation ox about any nominal design variable value. At higher values of x, the sensitivity of y is less, and so the variation about x results in less variation in the performance y. In this case, higher values of x are more robust.

A way to think about this circumstance in terms of product design is to consider the various factors that can change the value of performance. As shown in Figure 19.3, three different types of inputs determine the performance that any product will exhibit at any point in time. There is a nominal design that is selected; altering it changes the performance. This nominal design is the choice of the design team. There are also other factors, however; manufacturing variations and variations in material or environment can all change the performance level. These are noise variations. There are also other variations that are unique forms of "noise;' such as differences in how the operator uses the product. This type of noise is discussed in greater depth, later in the chapter.

The task of robust design is then to construct a product model including performance, design, and noise variables, and to then use the model to improve the design by selecting a design configuration that provides low performance deviations when the noise variables are free to vary. This robust selection process is depicted in Figure 19.4. There are basically three phases to the process. First, the model must be constructed. The performance, noise, and design variables must be identified. Next, the performance variation must be measured as the noise varies. This step must be repeated at different design configurations, to determine which design configuration has less performance variation, all for the same input noise variation. At each product configuration, this task involves taking a set of performance measurements and reducing them to a single rating for the design configuration. Lastly, the most robust configuration must be selected, based upon this robustness rating.




Date: 2016-03-03; view: 746


<== previous page | next page ==>
Plan for the lesson | Noise Variable Matrix
doclecture.net - lectures - 2014-2025 year. Copyright infringement or personal data (0.007 sec.)