Tuesday, October 30, 2012
STRATEGY TO DESIGN IN QUALITY
Methodologies to proactively assure high quality and reliability by design are taught in Dr. Anderson's in-house DFM training seminars and are published in the book, Design for Manufacturability & Concurrent Engineering, Copyright 8 2010 by Dr. David M. Anderson. Here are excerpts:
$ Observe quality and reliability design guidelines; 29 guidelines are presented in Chapter 10, A Design for Quality,@ in the book Design for Manufacturability & Concurrent Engineering.
$ Understand past quality problems. Thoroughly understand the root causes of quality problems on current and past products to prevent new product development from repeating past mistakes. This includes part selection, design aspects, processing, supplier selection, and so forth. It may be useful to have Manufacturing, Quality, and Field Service people make presentations to newly formed product development teams showing, hopefully with some real life examples, some of the past problems that can avoided in new designs.
$ Raise and resolve issues early by: learning from past quality problems; early research, experiments, and models; generate plan-B contingency plans; and proactively devising and implementing plans to resolve all issues early.
$ Use Multi-functional teamwork. Break down the walls between departments with multi-functional design teams (Deming's 9th point) to ensure that all quality issues are raised and resolved early and that quality is indeed treated as a primary design goal.
$ Utilize Quality function deployment (QFD) to define products to capture the voice of the customer the first time without the cost and risk of changing the design. QFD is one of the techniques in the collection of tools known as A Design for Six Sigma.@
$ Do thorough up-front work (a key element of Concurrent Engineering) so product development teams can optimize quality starting with the concept/architecture phase and avoid later quality and ramp problems.
$ Simplify the design for the fewest parts, interfaces, and process steps. Elegantly simple designs and uncomplicated processing result in inherently high quality products.
$ Minimize the exponential cumulative effect of part quality and quantity by specifying high-quality parts and simplifying the design with fewer parts. The formula and graph (at the right) state that the quality of the product (the first-pass accept rate) will be (assuming perfect processing) equal to the quality level of the parts to the exponent of the number of parts! So, for instance a product with 500 parts with each part being 99.9 percent good, a third of the products will fail just from the parts.1
High-quality parts can remove that variable from product development efforts, thus assuring that achieving the desired functionality will not be delayed – or incur more cost – because of part quality issues.
$ Select the highest quality processing. Automated processing produces better and more consistent quality than manual labor.
$ Optimize tolerances for a robust design using Taguchi MethodsTM to ensure the high quality by design. This is a systematic way to optimize tolerances to achieve high quality at low cost.2 It does this by using Design of Experiments to analyze the effect of all tolerances on functionality, quality, and manufacturability to analyze tolerance A stacks@ and A worse case@ situations. The procedure can identify critical dimensions that need tight tolerances and precision parts, which can then be toleranced methodically. The unique strength of this approach is that it can minimize cost while assuring high quality by identifying low demand dimensions that can have looser tolerances and cheaper parts.
Such a design would be considered robust so that it could be manufactured predictably with consistently high quality and perform adequately in all anticipated usage environments. It would also ensure that margins are adequate for current and future components from current and future suppliers.
Without a methodical way to determine tolerances, the alternatives would be: (1) make all tolerances tight A just to be sure,@ which is expensive. Tolerances that appear to be overly tight may have credibility problems and invite interpretation or (2) inadvertently (or deliberately) make tolerances too loose, leading to manufacturability and quality problems. Performance, quality, and manufacturability problems may be inconsistent and thus hard to troubleshoot and rectify. Robust design is one of the techniques in the collection of tools known as A Design for Six Sigma.@
$ Utilize Poka-Yoke principles applied to product design to prevent mistakes by design in addition to traditional manufacturing techniques to prevent incorrect assembly or fabrication
$ Proactively minimizing all types of risk, not just functionality. For critical applications, use Failure Modes Effects Analysis (FMEA), which is one of the techniques in the collection of tools known as A Design for Six Sigma.@
$ Base metrics and compensation on Total Cost and Time to Stable Production to avoid compromising quality with cheap parts to save A cost@ or throwing a sub-optimal design over the wall A on time.@ In many organizations, individuals do what they are rewarded to do. If they are rewarded for releasing a design A on time,@ they will, effectively, throw it over the wall on time, ready of not! If they are rewarded for achieving A cost targets@ without total cost measurements, they will do so by buying the cheapest parts available, probably without concern for part quality. So reward systems must be structured to include quality metrics
$ Reusing proven designs, parts, modules, software objects, and processes to minimize risk and assure quality, especially on critical aspects of the design.
$ Document thoroughly and completely. In the rush to develop products, many designers fail to document every aspect of the design thoroughly. Drawings, manufacturing instructions, and bills-of-material sent to the manufacturing or vendors need to convey the design unambiguously for manufacture, tooling, and inspection. Imprecise drawings invite misunderstandings and interpretation, which add cost, waste time, and may compromise quality. Centralize the most current data with good product data management.
$ Thoroughly design the product right the first time. Use Design for Manufacturability techniques presented herein to ensure that the product is design right the first time. If quality is not assured by the initial design, then expensive change orders will have to be carried out, wasting valuable engineering resources and possibly inducing further quality problems in the process. Be sure to be able to comfortably satisfy design goals and constraints without having to compromising the product just to get products out the door.
Designing for quality is what gets quality from 5 sigma to 6 sigma.
Posted by Joao Moraes at 11:02 PM