| QG14: Tolerance Design P Diagrams |
|
Once a design is made robust to noise using set points developed from both analytical & empirical methods (typically Modeling & Simulation and various DOEs), it is time to explore just how much the "robust set points" can be allowed to vary in production and life-cycle maintenance of the product and its production systems. This is a form of engineering analysis that is often done at a cursory level, leaving the specified tolerances in a state of "under-development". This is especially true when one asks whether the tolerance ranges were developed under stressful noise conditions. The usual result is high-cost, hard-to-hold tolerances that have additional quality issues under stressful application conditions in the customer's environment.
.jpg) | | Tolerance Parameter Diagram |
It is important to note that both target set point adjustments, as well as tolerance range adjustments, can be evaluated to dial a design "in" to identify the most economical set points (both target set points as well at their tolerance ranges!). By Target (T) and tolerance ranges we mean (T value +/-tolerances). Sometimes you can readjust Target set points to gain the ability to loosen otherwise extremely tight / expensive tolerances on a particular parameter. This may change the robustness a little but it may be a trade-off worth making. As always, if a tolerance range affects mean functional performance when it is exercised, the mean of Y can often easily be adjusted on to the desired target using the Critical Mean Adjustment Parameter(s). These are often defined by Response Surface Methods from the DOE toolset and used in production adjustments during assembly.
Once a Tolerance Parameter Diagram is fully illustrated, the Team constructs a special form of DOE that will explore tolerance set point ranges to identify optimal tolerances under nominal and, if desired, under stressful conditions (known as "robust tolerances").
These diagrams serve two useful purposes: 1) They are essential to constructing a coherent Design Guide so people understand how the design tolerances work and why they were set where they are, and 2) they set the stage for conducting Tolerance Analysis for Modeling & Simulation and Designed Experimentation (DOEs).
|
| QG15: The Big 7 Metrics Overview |
|
There are 7 major metrics, we call them The Big 7, that allow you to know if you have the data that is reasonable proof that a candidate NUD Y or X parameter is critical or key or just merely important and deserving of ECO classification:
- Measurable:
Is the function or part/material characteristic able to be measured? Do you have a capable measurement system and an acceptable Gage R&R Study to prove it? - Stable:
Is the function or characteristic repeat ably stable when you take sequential sample data over time? Is it under statistical control with an SPC (I&MR) chart to prove it? - Adjustable:
Is the function or characteristic able to be consistently and repeat ably adjusted over a desirable range? Can you document this range with a regression plot and its equation with an acceptable Coefficient of Determination (R2 ratio value for the Model)? - Independence, Interactivity and Statistical Significance:
Are the Xs that control Y statistically significant, independent of one another or are they interactive? Can you provide a p value and F Ratio for each independent X and those that are interactive with one another? - Sensitivity:
Are there significantly strong sensitivities (Y/X) that exist between specific X-Y relationships that are hyper-sensitive and problematic? Do these sensitivities present extremely tight and high cost tolerance requirements? - Robustness:
Are there reasonably strong interactive relationships between controllable Xs (XControl) and Noise Factors (Xnoise)? Can you prove with data that certain set points for the candidate controllable X parameters leave the Y more insensitive to Xnoise parameters than others? Which Xs have a dominant effect on the standard deviation of the Ys when unwanted noise parameters are changed? Can you adjust the standard deviation of the Ys? - Capable:
Can you demonstrate Cp and Cpk values for each candidate Critical or Key Y and X as you mature the sub-level and system level functions? Can you show growth in the Cp and Cpk values over time (phase-by-phase) as you develop and integrate the system?
These 7 metrics will help keep you from being surprised if you conduct the tasks that produce the data for them at the earliest points possible in your development process. If you have an existing design and lack the parametric knowledge to make necessary changes in it, you will need to use these 7 metrics as you back-derive the parametric knowledge and Design Guide for future life-cycle management of this product and its production processes.
|
| |
Is there a topic you'd like us to write about? Have a question? We appreciate your feedback and suggestions! Simply "reply-to" this email. Thank you!
Sincerely,
|
Carol Biesemeyer
Business Manager and Newsletter Editor Product Development Systems & Solutions Inc. |
