Quality Scan: The Great Disconnect Between Design, Manufacturing, and Measurement
By Edward S. Roth, PE, FSME, LSME, CMfgE
Past President (1977-78)
Society of Manufacturing Engineers
After 1982, the American Society of Mechanical Engineers (ASME) took over control of the Y.14M drafting standard from the American National Standards Institute. This dimensioning and tolerancing standard provides ASME with a prime source of income. The first disconnect between design, manufacturing, and measurement occurs because the standard is only taught in community colleges and other two-year institutions, and not in mechanical engineering programs at four-year universities.
Because mechanical engineers are not exposed to ASME Y14.5M-2009, they delegate the application of the standard to drafting by layouts, sketches, and verbal instructions, which initiates the second disconnect. As neither design engineers nor manufacturing engineers are involved in the drafting, engineering drawings are thrown over the wall to manufacturing with no input from the engineer who makes the product—creating the third disconnect. Manufacturing engineers receive engineering drawings that can contain as many as 44 unrelated geometric tolerances and datums. This flies in the face of convention, as manufacturing engineers fixture parts so manufacturing operations can be competed in one setup. This enables the achievement of the best interrelated geometry, so that all features can be created equally.
The fourth disconnect occurs during measurement activities. The use of CMMs—both portable systems and larger machines with 3-D software that can be based on the manufacturing floor—is increasing. Both types of machines can measure a part in a single setup, or measure it in real time in-process using a touch probe in the machine tool. All these CMMs, whether small units or systems large enough to measure an 18-wheeler, rely on the use of a ball on the end of a stylus. The ball scans the surface of a feature, which can have any geometric form—flat, curved, tapered, cylindrical, etc. These computer-driven measurement systems can also generate 3-D functional gages and determine if they mate with the computer-generated part. Because these "soft gages" should represent the worst-case mating part, creating them becomes almost impossible because of all the unrelated geometric tolerances based on unrelated datum references. As an example, the figure on page 60 of SME’s Aerospace & Defense Manufacturing 2010 Yearbook shows five totally unrelated datums and seven geometric tolerances, resulting in many stackups. No manufacturing engineer worth his salt will do what this drawing demands. Instead, he or she will look for a way to achieve a single setup by "blind siding" the part if at all possible. My guess at an approach to the part would be: Use the bottom surface as primary Datum A; the far side to bank against as secondary Datum B; and the back end as tertiary Datum C. These three surfaces will be machined first so they will be perfectly related, then fixtured so the rest of the part features can be produced in one setup.
So we have an interesting set of circumstances here. The ASME Y14.5 standard has hundreds of complex callouts, all defining solid features, and the dominant measurement system is a CMM that moves a ball-tipped stylus across a surface, taking measurements along the profile of a line.
How do we define a product so that the engineering requirements are satisfied by the profile of a line? One apparent solution is the activity of an ASME standards committee attempting to mathematize GD&T. I wish them luck when it comes to composite tolerancing, which flies in the face of single-setup manufacturing even the best quality-control activists are unable to chart.
We wouldn’t dare call out the profile of a line to define all curved, flat, and cylindrical features, which is exactly what CMMs and laser line scanners actually do—but that is how they are currently measured. Perhaps calling out the profile of a surface for flat, curved, or cylindrical surfaces, which is assumed if the CMM covers a sufficient number of lines on the feature, will work—although the number and spacing of the lines, and the number of points recorded will vary with the CMM and the operator. Indeed, this approach demands that someone define the spacing and frequency of the paths. What do we do? The readers of the article have a choice to either accept these disconnects as-is, or to contact ASME and the committee personnel. Doing nothing certainly will influence our nation’s wealth-generating capability—in a negative way. ME
This article was first published in the June 2011 edition of Manufacturing Engineering magazine. Click here for PDF.