Featured Product
This Week in Quality Digest Live
Metrology Features
Prashant Kapadia
Aid root cause analysis by detecting wear and tear in advance
Mike John
How metrology can bring the medical-device design and manufacturing process together
Steven Brown
21st-century standard candles at NIST
It’s actually the differences in a twin that are most useful

More Features

Metrology News
On-demand pipe flow measurement, no process interruptions
Q-Span Gauging Systems improve efficiency, capacity, and real-time reporting of quality departments
Products range from software to scanners
See MobileCollect in Action in Booth No. 135413
Four new models available in TVM Series
Leader in 3D scanning joins master metrology users at this unique event
At Vision Spectra Conference, July 19, from 7:30–8:00 a.m. EDT

More News

William Tandler


The GD&T Encoding Process

Smart GD&T workshop No. 7

Published: Wednesday, April 9, 2008 - 22:00


he actual purpose of GD&T:

It’s common to think that the primary purpose of GD&T is to unambiguously communicate design intent to manufacturing and inspection. In fact, the most important objective is to ensure that what we communicate is worth communicating, and that it represents a part that can be guaranteed to assemble and operate, if the specified tolerances are met. Bad GD&T—the decorative kind—however, is very dangerous. Machine shops always try to manufacture exactly what we say we want, but if what we say we want isn’t what we actually need, they’re generally held responsible, often not rightfully, and we all suffer.

The actual purpose of GD&T is therefore to encode the function of each feature of a part by specifying limits of dimensional imperfection that guarantee its operation and assembly, and at the same time minimize manufacturing and assembly costs. In fact, the GD&T encoding process addresses six sometimes competing functions. These are listed in figure 1 in the order of their importance. As also indicated, it is not only the functions themselves that are important, but also the forces that act as the functions are performed.

Figure 1

Encodable Feature Functions

1. Operation


2. Assembly


3. Cosmetics


4. Manufacturing


5. Gauging


6. Metrology


Clearly, the most important function is operation, because if a part is manufacturable and assemblable, but won’t operate, why go to the trouble? The next most important function is assembly, because if the part won’t assemble, again, why manufacture it? Next comes cosmetics, because if things don’t look good, customers won’t buy. Next comes manufacturing, which we need to favor to keep costs in line, but surely not at the expense of assembly and operation. Then comes gauging, or functional gauging, which we might make easier and cheaper through the use of the modifier (M) in place of (S), but, again, never at the expense of reliable assembly and operation. Finally, at the bottom of the list, we come to coordinate metrology, and here we have to say: “If we can make it, it looks great, assembly’s no problem and the thing operates like a charm, but you can’t measure it, then, in general, metrology has a problem—and they need better tools.” In general, as we go down the list, we must try to make everyone happy—happiness saves money—but only if it’s not at the cost of a higher-precedence function.

Encoding process preparation
If we’re to succeed with GD&T, we need to proceed in an orderly fashion. The Smart GD&T approach is set forth in figure 2. Step 1 requires us to do a feature inventory. This helps us get a first feel for the part. Step 2 requires us to analyze the function of the part as a whole in keeping with the items in figure 1, in particular how it interacts with mating parts during operation, assembly, manufacturing, and gauging.

Figure 2

Smart GD&T Encoding Process Steps

1. Create a feature inventory for the selected part.

2. Analyze the function of the part.

3. Determine the function of each feature and create a feature hierarchy.

4. Encode the function of each feature of the part by determining what can go wrong, and then partially “stuffing” the necessary feature control frames step by step, to begin to impose limits.

5. Finish the feature control frame “stuffing” process by selecting tolerance values and the proper tolerance zone size and mobility modifiers.

6. Decode the initial feature-control frame set to determine if the functions they assign to each feature are the most effective for the overall function of the part. If not, modify the code to represent the better understood, newly assigned functions.

Step 3 requires us to dig a little deeper to determine the function of every feature of the part. Once complete, we can establish a feature hierarchy: Determine the most important feature, then the next most important, and so on. This hierarchy determines the geometry control chain in which each feature is linked to every other feature and prepares us for step 4, which requires us to encode the functions we have discovered. This is the point at which we need to proceed carefully, and occasionally even make changes in the CAD model if we discover missing features or shortcomings in a feature’s functionality.

Step 6 requires us to decode the code to discover what we have actually said, and to determine the effects of what we have said on all six functions listed in figure 1. If in the process we discover better ways to achieve the now better-understood functions, it’s time to reencode. The overall encoding process is therefore iterative, and the more we engage the people who really understand what’s going on such as the machinists, the inspectors, and the assembly crew, the sooner we will truly latch and complete the code getting it right.

An example
We’ll illustrate the Smart GD&T encoding process in figure 2 using the simple flange shown in figure 3. Here we go:

Step 1
Create a feature inventory: As currently modeled, the flange consists of two planar surfaces and eight cylindrical surfaces.

Step 2
Analyze part function: The function of the flange is as follows: When the flange is mounted to the mating flange with six bolts threaded into mating bores, the objectives are 1) for the mating planar surface to create a reliable pressure seal against a rubber O-ring, and 2) for the central bore in the flange to be coaxial with a central bore in the mating flange.

Step 3
Create a feature hierarchy: The most important feature of a part is the one that constrains the most important degrees of freedom—pitch and yaw—during the assembly process. The most capable feature in our case is the face of the flange that will engage the mating flange. Because there are two faces and each will require different tolerances, it will be necessary to destroy the symmetry of the flange as now shown in figure 3, a move we’ll make when we start “decorating” the model. The next most important feature is the one that constrains the next most important degrees of freedom during the assembly process, in this case the mutual coaxiality of the flanges.

Figure 3. The nominally dimensioned CAD model

Barring any unusual intervention (to be discussed later) the features naturally responsible for this are all six bolt holes. With all necessary degrees of freedom now constrained, the next most important feature is surely the central bore, followed by the periphery of the flange and the second planar surface.

Step 4
Feature-control frame stuffing: We are ready to start “decorating” our CAD model.

First planar surface: The first thing we realize is that there are two planar surfaces, and we’ll need to differentiate between them. To do so, we destroy the symmetry of the flange by placing a highly visible chamfer on the less important face. Next, as shown in figure 4, we identify the most important face by giving it a datum feature label. Next we ask “What could go wrong?” for example, “Could there be a problem with size?” No, planar surfaces have no size. “Could there be a problem with form?” Yes, the face might be warped, which would cause a sealing problem. “Could there be problems with orientation and location?” No, because there are as yet no other features relative to which we could control orientation and location. We address the only potential problem with a flatness control, which, for the time being, lacks a tolerance, because it’s much too early to be able to select one.

Figure 4. Starting the feature-control frame-stuffing process

The bolt hole pattern: Next we consider the six bolt holes, and again ask “What could go wrong?” In this case everything—we must deal with their size, their form, their mutual orientation, and their location, as well as their perpendicularity to datum feature A. We cover size and form using the diameter tool, but again leave the tolerance undetermined. We control mutual orientation and location with the position tool, and, by including a reference to A, limit their perpendicularity to A as well. We again leave the tolerance value to be determined later, and note the need to select a tolerance zone size (TZS) modifier at the appropriate time. Because only basic dimensions can orient and locate tolerance zones, we add an “equally spaced” note to deal with the required basic angles, and place the bolt hole pattern radius in a rectangular frame. Finally, we identify all six bores as datum feature B by attaching a label to their feature control frame. See figure 4 for details.

The central bore: Here, size, form, orientation, and location again need to be controlled relative to a datum reference frame established using planar surface A and bore pattern B. We select the appropriate size and position tools, and again leave all tolerance values to be determined later, as shown in figure 4.

The periphery of the flange: Again all four variables must be controlled, and we again choose the typical tools, size, and position, and leave tolerance sizes and modifiers for step 5.

The opposing face: In the case of this last feature, we realize that we will need to loosely control its flatness as well as its parallelism and location relative to A. We decide to do so with the good old fashioned “width” tool.

Step 5
It’s time to finish the stuffing process by filling in the blanks.

First planar surface: After talking to our friends in the shop about their ability to hold flatness on a turned face, and after considering the effect on the sealing function, we select an easily achievable flatness tolerance of 0.01 mm. See figure 5.

The bolt hole pattern: If the bolt hole patterns in both parts are to be responsible for the coaxiality of the mating central bores, we’ll have to control all their characteristics quite tightly. Based on manufacturing cost data, we select affordable size and position tolerances which also meet operational and assembly requirements. Next we must select the tolerance zone size (TZS) modifier for position that best encodes our objectives.

Figure 5. Feature-Control Frame-Stuffing Refinements

Although the fundamental function of bolt holes is clearance, which would point to a TZS modifier (M), these bolt holes have a critical locating function that will suffer as they grow larger. To avoid encouraging the shop to go for the least material condition, the natural consequence of an (M), we therefore make our TZS modifier (S). Finally, we add chamfers to the bolt holes to ensure clean and sturdy edges. See figure 5 for details.

Manufacturing note: Of course, when our friendly machinists see the drawing, they complain about making the bolt hole pattern a datum feature (what we call a composite datum feature), but our suggestion that they will surely invent a temporary, manufacturing datum feature for their purposes (the outer periphery of the flange?) gets them to go along with our plan. Inspection note: The coordinate metrologists also complain about the use of a bolt hole pattern as a datum feature, citing the trouble patterns cause their software, but we know that they have workarounds and stick with our decision (Are we having fun yet?).

The central bore: Because coaxiality is the primary functional objective, and size, as we now determine, is not critical, we select a loose size tolerance and add a cylindricity control and a still-affordable but tight enough tolerance for position. Now, if coaxiality is as important as we understand it to be, surely no benefit can be had from allowing the position tolerance to grow as the central bore grows. As a result, we select the TZS modifier (S). Finally, since there also is no benefit in allowing the central bore to be more offset as the bolt holes grow in size (the datum shift question), we absolutely must select (S) for the tolerance zone mobility (TZM) modifier for datum feature B. As a last step, we add a “break edges” requirement, and trust that our suppliers know what we mean. All this is shown in figure 5.

The outer periphery of the flange: Here again, all four variables—size, form, orientation, and location—must be controlled, and we choose the typical tools as shown in figure 5, but this time with very loose tolerances. We could of course make both TZS and TZM modifiers (M), but the specified position tolerance is so large as to offer no manufacturing cost benefit, and cosmetics concerns dictate against encouraging sloppy work here. Again, we add a “break edges” requirement to the back-side of the flange to complete control of the feature.

The opposing face: We decide to stick with the good old fashioned “width” tool here, and use a rather large tolerance—perhaps unnecessarily large, but, we like to be friendly when we can—because variations in thickness can be compensated for by the mating bolts. However, we decide to impose a still affordable refining tolerance on the parallelism, to ensure that the big width tolerance doesn’t result in a parallelism error which could put undue strain on the bolt heads. See figure 5.

Step 6
Tired of stuffing feature-control frames? Good. It’s time for some decoding to analyze our stuffing. First planar surface: All is well here. The mating face of the flange is definitely responsible for constraining pitch and yaw, and is therefore definitely the primary datum feature. See figure 6.

The bolt hole pattern: After much back-and-forth, we come upon the bright idea that we could probably devise a simple fixture to make the outer peripheries of both flanges coaxial during the assembly process as the basis for ensuring coaxiality of the central bores. To encode that alternative, we make the outer periphery datum feature B instead of the bolt hole pattern. (See figure 6.) Because the function of the bolt hole pattern is now purely to provide clearance for the mating bolts, we can significantly enlarge the size and position tolerances, and replace the TZS modifier (S) with the more forgiving alternative (M). However, the TZM modifier on B must remain (S), because changes in the size of datum feature B bring no benefit, because it is dedicated to constraining coaxiality.

As might be suspected, these changes earn us congratulations from the machine shop, the quality assurance team, and the assembly team, as well as from management, based on their contributions to the bottom line.

Figure 6. Turning the code upside-down for a final landing

The central bore: No changes are necessary here, except the reference to the new datum feature B, which is automatic. See figure 6.

The periphery of the flange: Because this feature now has centering responsibilities, it must be more tightly controlled, and based on the intended manufacturing processes, these tighter controls are easy to meet and will add no cost.

The opposing face: No changes. Hooray!

Last steps
Of course we are far from finished, because we have paid no attention to the mating flange, and the effect on it of our controls on the flange at hand. Oh well, stuff for another day...

Future articles
In our next Smart GD&T Technology workshop we’ll complete the encoding process on the mating flange and do a tolerance stack-up analysis (a supah TSUPA) to assess the worst possible coaxiality of the two central bores. In workshop No. 9 we’ll investigate the effect of GD&T on manufacturing, and in workshop No. 10, dive into the coordinate metrology aspects of the game.

Feedback on topics of interest
We are very grateful for e-mail feedback on workshops No.5 and No.6 from Carlos Gonzalez Gonzalez and from Brian Heersink. Please keep your questions and suggestions coming.

By the way, if you feel that this might be of interest to your design crew, why not forward a link to see what they think of it?


About The Author

William Tandler’s picture

William Tandler

William Tandler’s professional experience includes three years at the Hewlett Packard R&D laboratory in Palo Alto, CA, where he helped develop a quadruple mass spectrometer; five years with the laser manufacturer Coherent Inc. as international marketing manager; followed in 1975 by the founding of Multi Metrics Inc.

Tandler is currently a member of the ASME Y14.5.1 Mathematization Committee; a subject matter expert for Working Group 4 of the ASME Y14.5 standard focused on datums; and has been a contributor to ISO TC213 in the realm of datum reference frame construction.