1. Robust Design Chapter Six

2. Training Manual January 30, 2001 Inventory #001449 6-2 6. Robust Design In this section, we will present a practical application of optimization tools - using Robust Design concepts for mass production. The following topics will be covered: A.Define robust design B.Using optimization tools to achieve a robust design C.A workshop demo and/or exercise

3. Training Manual January 30, 2001 Inventory #001449 6-3 Robust Design A. Definition What is Robust Design? A design that is not sensitive to variations in noise parameters such as: –manufacturing tolerances –material properties –environmental conditions - temperature, humidity, etc. I.e, a design that is robust!

4. Training Manual January 30, 2001 Inventory #001449 6-4 * Quality Engineering Using Robust Design, Madhav S. Phadke, Prentice Hall, 1989 Robust Design...Definition Robust design concepts are especially useful for mass production, where minimizing the variation in the product characteristics is just as important as obtaining the optimum design. In a study*, SONY television sets were manufactured in U.S. and Japan to identical designs and tolerances. –The U.S. factory used the fraction-defective criterion. –The Japanese factory used the Robust Design concept. –Japan produced more sets near target color density and were preferred by U.S. consumers.

5. Training Manual January 30, 2001 Inventory #001449 6-5 Robust Design B. How to Implement ANSYS optimization tools - Gradient, Factorial, and Sweep - can be used effectively to implement robust design concepts. One way is as follows: 1.Identify noise parameters and control parameters. 2.Create a parametric model using noise and control parameters. 3.Determine which noise parameters have the most effect on the design. 4.Determine which control parameters can be changed to reduce the effect of the noise parameters.

6. Training Manual January 30, 2001 Inventory #001449 6-6 Robust Design...How to Implement 1. Identify Noise and Control Parameters Noise Parameters are factors which can vary randomly and are beyond the control of the designer, e.g: –Material properties –Manufacturing tolerances –Environment, e.g, temperature and humidity –Component degradation with time Control Parameters are factors that the designer can change to decrease the effect of the noise parameters.

7. Training Manual January 30, 2001 Inventory #001449 6-7 Robust Design...How to Implement 2. Create Parametric Model Use both noise parameters and control parameters to build the parametric model. Specify noise parameters and control parameters as DVs. Remember that DVs can only take on positive values. For example, if a dimension thk = 2.5 has a manufacturing tolerance of thktol =  0.001, you would build the model using thk+thktol as the thickness dimension: thk = 2.499 tol = 0.002

8. Training Manual January 30, 2001 Inventory #001449 6-8 Robust Design...How to Implement 3. Determine Noise Parameters That Have the Most Effect The obvious tool for this is the Gradient Tool, which shows how a given design (the reference design) would respond to a 1% change in each DV. The reference design in this case would be the current, accepted design. The curve with the steepest slope indicates the noise parameter that has the most effect on the design.

9. Training Manual January 30, 2001 Inventory #001449 6-9 Robust Design...How to Implement At this point, one option is to somehow reduce the variability of the critical noise parameters. For example, if a tolerance is deemed critical, you might be able to tighten the manufacturing tolerance. However, most noise parameters, by definition, are beyond the designer’s control. The other option is to move to step 4 to investigate whether the effect of these noise parameters can be reduced by changing the control parameters.

10. Training Manual January 30, 2001 Inventory #001449 6-10 Robust Design...How to Implement 4. Determine Which Control Parameters Can Be Changed The Factorial Tool is the obvious choice for this. It can be used to determine two-way and three-way interactions between noise parameters and control parameters. For example, what value of THK will make THKTOL least effective? Another way is to use the Sweep Tool to sweep through several control parameter values for a given noise parameter value.

11. Training Manual January 30, 2001 Inventory #001449 6-11 Robust Design...How to Implement If you determine that a different value of a control parameter, say THK, does indeed reduce the effectiveness of a noise parameter, you will need to make sure that the new design still performs to desired standards. This is just a brief introduction to basic Robust Design concepts and how ANSYS optimization tools can be used to implement them. For more information, please refer to the Appendix in your Workshop Supplement entitled: World Class Quality through Robust Design.

12. Training Manual January 30, 2001 Inventory #001449 6-12 See your Design Optimization Workshop Supplement for details. Robust Design Workshop In this workshop, we will do a factorial analysis of a magnetic actuator.