Geometric Dimensioning and Tolerancing (GD&T) has been around since the 1940s, yet it is unappreciated as a way of finding bad designs before parts get made.
Edited by Leland Teschler
Geometric Dimensioning and Tolerancing is a well-established enabling technology, so you might think it would be widely used. But in fact, few managers understand how valuable this technology can be. GD&T can’t make a bad design good, but the proper use of GD&T within a well-defined product development effort can identify a bad design before it results in a lot of needlessly expensive parts.
There are two components to the technology of GD&T. First, it is a precise communications tool. It uses a set of symbols, rules, and definitions to mathematically define part requirements. Second, it is a design approach that lets the engineer define parts based on customer requirements and part functions while allowing maximum tolerances for manufacturing. This combination, properly executed, results in high quality and low costs.
GD&T is part of a larger effort, a product-development process (PDP). Some companies may not even realize they have such a thing. But they do execute a series of steps in the process of developing a new product and bringing it to market. Usually, the more complex the product, the more defined the PDP.
A PDP serves many purposes. First, it defines a series of activities that translate new-product concepts into customer requirements that drive engineering designs and testing. Second, a PDP distributes and harmonizes activities between different departments (marketing, purchasing, engineering/design and manufacturing). Third, a PDP provides a plan for all these activities.
The Myths of GD&T
Myth: We don’t need GD&T because our drawings are good enough without it. Reality: Without GD&T you cannot accurately create tolerance stacks. You cannot accurately inspect your parts. The dimensioning cannot represent the product requirements.
Myth: GD&T is confusing — everyone has a different interpretation. Reality: There is a kernel of truth in the statement, but it is not the fault of the language. Incomplete or poor tolerancing on drawings is subject to more than one interpretation. Untrained drawing users may feel there are multiple interpretations. Proper drawings and a skilled workforce can greatly reduce this problem.
Myth: It takes longer to apply GD&T and we don’t have the time. Reality: If your engineers have the right skill set, it is quicker to specify symbols in place of lengthy notes.
Myth: Our suppliers don’t understand GD&T so we don’t use it. Reality: Why would you change your drawings to have less tolerance and be less clear to accommodate unskilled suppliers? Would you stop using e-mail or solid models if your suppliers didn’t have these capabilities? The use of these technologies became requirements for suppliers. GD&T has been around for many years. Require your suppliers to understand drawings.
Myth: Using GD&T raises part costs. Reality: When properly specified, GD&T has a number of tools that provide larger tolerances to reduce part-manufacturing costs. To name just a few: Round tolerance zones, Bonus tolerance, Rule #1, Composite tolerances, Functional dimensioning, Separate requirements.
Myth: With solid models, I don’t need GD&T. Reality: This may become true someday, but for now, we still need tolerances to functionally describe part features, allow maximum tolerances, analyze new designs, analyze production or field problems, and to inspect parts.
In short, a good PDP reduces the time it takes to deliver a quality product to market. So it is useful to review the role GD&T plays within a PDP. A typical PDP consist of the following stages:
• Requirements setting
• System/conceptual design
• Component/detailed design
• Manufacturing design
• Component validation
• System validation
• Manufacturing validation
Product variation from manufacturing is a fact of life. The allowable variation is communicated through a GD&T feature control frame. The Quality Management Process focuses on how this variation is to be controlled and managed. It is useful to examine a GD&T feature control frame and illustrate how it is used through the PDP. The consequences of not using GD&T in a PDP can be costly.
Datum selection is the best place to start when managing variation during the PDP. A robust datum scheme uses datums that best prevent the parts from moving out of position, minimizes the number of items in a tolerance stack, is well controlled within the part (location tolerances and rigid features) and is shared by as many manufacturing processes as possible.
The datum scheme is largely driven by the product-build strategy and system requirements: Thus it is important to consider the effect of datum selection early in a PDP (requirements and system design phases). The design-and-build strategy defined at this stage will have the largest effect on product quality with the least cost. Mistakes made at this stage will be expensive to fix at later stages, if they can be corrected at all.
The requirements-setting phase of a PDP is where you identify product features and performance requirements which dictate what needs to be controlled and how tight to hold the part tolerances. The part-appearance requirements dictate where there are critical dimensions, as well. Remember, the best datum reference frame is only relative to the specific product features that must be controlled.
You must select functional datums. A functional datum is simply one that uses the product features that physically locate the part to the final product. Using any other datum will add variation in the final tolerance stack up.
Also remember the importance of datum priority. The primary datum should be the functional datum that controls most of the allowable degrees-of-freedom of movement. The secondary and tertiary datums, if needed, are functional datums that control succeedingly fewer degrees of freedom. During the design of the manufacturing process, you must select features used to locate the part during manufacturing operations — features that determine where and how the part is held while it is manufactured. These features must be either identical to the product datums or extremely close to them. When locating features don’t coincide with functional datums, variations invariably arise.
Finally, by definition, the validation phase of a PDP employs the GD&T datums in all fixtures used for checking dimensions, or for CMM (coordinate-measuring machine) routines that gauge part variation. You should check fixture designs and CMM routines to ensure they locate parts to the specified GD&T datums and conditions (restrained or unrestrained). All in all, you’ll want to identify functional datums and plan for GD&T early in a PDP to maximize part quality.
Symbols and tolerances
With product features and datum references settled, it’s usually a simple matter to choose a geometric characteristic symbol. The type of product feature to be controlled and the type of control (form, profile, orientation, location, or runout) will usually narrow the field to one or two symbols. The geometric feature being controlled is directly related to customer requirements. ASME Y14.5M1994 is a good reference for a complete description of each symbol and its use.
One of the largest mistakes made in product design is copying tolerances from a previous drawing without reevaluating how well individual part tolerances and the datum scheme meet product requirements. Too often we have heard “…but the part is within specification,” when the part tolerances specified were never evaluated to see if their combination met product requirements.
Using CAD Data in place of a toleranced drawing
That said, readers should understand that CAD models do indeed play a valuable role in the PDP. Many companies successfully produce prototype parts from models without tolerances. There are certainly many benefits to using solid models in a PDP. Nevertheless, the process can be a minefield if not managed properly, as the following story shows. This is really about several companies across several industries, but for simplicity, I will explain the events as if they happened at one organization.
These companies all wanted to shorten the development time and reduce product costs while ensuring a quality product. Each firm convinced themselves that by using models-only without tolerances, they would save time by not having to make drawings or do a tolerance analysis. They aimed to verify assembly and clearances using the models. They also wanted to procure the prototype parts directly from the solid models and thus have more time for development testing. They intended to make the production tool right from the model on long-lead items and order it as soon as development testing was far enough along to ensure a stable part configuration. All in all, they figured on saving 12 to 18 months in their PDP process.
Things started out quite well. The parts were coming in early, and testing and debugging proceeded normally. They were about 12 months ahead of the traditional method of using 2D drawings. Then the problems started.
It was diffcult to get quotes for production. Suppliers were replying with “No quotes” because they needed dimensions and tolerances to know if the parts were acceptable. The fact that the tools were made from the models was not an assurance that the part variation would produce an acceptable part.
Where parts were produced from the untoleranced part design, manufacturing had no idea of which dimensions were sensitive to the product function and which were not. The organization assumed there was no need for inspection or that they only needed to visually match the part to the model.
They eventually mapped the part, which was molded, and overlaid it with the 3D model. When the surfaces didn’t match perfectly, someone had to make the judgment call of whether the part was “close enough.” Several years and failed parts later, customer problems arose. There was no way to determine the size of the parts that were failing.
All of these manufacturers had emphasized solid models for several years. So many of their engineers and designers were not skilled or trained in tolerancing parts. When engineers and designers don’t need to account for part tolerances, they think in terms of a perfect model. But the parts that are produced contain variations that someone must account for.
Once our firm got involved, we helped assign tolerances to the individual part drawings, and then showed designers how to analyze the tolerances to ensure products would assemble and function. This process uncovered many additional problems and numerous tolerance stacks.
The bottom line: These manufacturers produced a design that would not assemble and couldn’t maintain critical clearances. The nominal design would fit together but could neither be inspected for use nor be used to fix a customer problem.
Each company learned three important lessons: First, tolerances are a vital part of engineering a product, even one described with a solid model. Second, engineers and designers need the skills to assign functional tolerances. And finally, the shortest path to production is one that considers the effects of tolerances early in the design process.
Tolerances should be allocated to each part based on product system requirements and manufacturing capability. Several methods exist to allocate tolerances. These are the same tools used for tolerance analysis (Monte-Carlo simulations, root sum of squares and limit stacks). In each of these methods the relative contribution of part tolerances to the final system variation can be calculated. Part tolerances need to be balanced against product and manufacturing costs. If system requirements cannot be met, first try reducing the part tolerance that contributes the most to system variation. If the cost of reducing a tolerance is high, try reducing other less costly tolerances instead.
Tolerance analysis must include part-fixture tolerances when a fixture-build is used in manufacturing. Otherwise the final tolerance analysis doesn’t consider the full range of variation. Note that use of GD&T allows additional tolerance for manufacturing (see GD&T Myths sidebar).
Of course, most companies are not building completely new products in which they have no prior experience. So they can use past data to help allocate requirements. It is helpful to look at the build strategies of past products and the resulting dimensional capability. This exercise will help show whether current requirements are in the realm of prior experience or if new techniques are in order. Ideally, your company has kept data on past part-tolerance capabilities. These data are not only necessary for continuous improvement, but also are useful early in a PDP. They help determine your process capabilities and what tolerances you must hold for specific manufacturing processes.
Tolerance analysis should be performed whenever build strategies or part tolerances change. This analysis is the final chance to predict and avoid problems before finalizing the part GD&T and ordering hardware.
Benefits of GD&T
One benefit of using GD&T is its ability to precisely and clearly document the part requirements. This results in a part that can be outsourced to anywhere on the globe. The mathematical precision of the part description makes the part easier to manufacture and inspect. Another benefit of GD&T is the ability to define a part in a manner that protects the part function and allows maximum tolerances for manufacturing.
GD&T is also vital for accurate results in inspection. The datum system communicates what part surfaces are used to create the datums for measurement. The feature-control frame communicates the sequence for relating the datums to the part surfaces. It also gives the amount of tolerance that part features are allowed from their theoretically exact location at the datums.
GD&T is important for calculating tolerance analysis accurately. The mathematical definition of part surfaces and their tolerance zones allows calculation and analysis of the extreme boundaries. This lets you analyze the effect of manufacturing variation before the part is produced, and thus anticipate problems and address them before the design is released.
Of course to reap the benefits, designers must properly specify GD&T on the drawing. In the rush to release drawings, sometimes engineers do not spend enough time on specifying GD&T correctly. The amount saved rushing through the drawing is miniscule compared to the potential savings from defining the part correctly.
There are a few practices that can help a company get the maximum benefit from GD&T. First and foremost, a company should create a policy mandating the use of GD&T according to a recognized standard, such as ASME Y14.5M1994. GD&T use must be consistent among engineering, design, manufacturing, and quality organizations. And all levels of management must endorse the use of GD&T.
Second, all portions of the organization that will be exposed to GD&T should be trained in it. The degree of training will vary depending on how the employees use GD&T (awareness overview, interpretation or application, analysis). Ideally, there should be a certification process. It should assess each employee’s knowledge of GD&T or need for further training and set requirements for each job’s level of certification.