In the realm of additive manufacturing, what exactly are the distinctions between accuracy, repeatability, and resolution? Accuracy describes how closely a manufacturing machine’s output conforms to a tolerance within a specified dimensional range. Repeatability captures the equipment’s capability to produce consistent output, time after time. And resolution refers to the smallest measurement the machine can reproduce.
These concepts are second nature to designers and manufacturing engineers. Yet, there is a widely held misconception about these measurement concepts as they relate to additive manufacturing. Over the years, some engineers have slid into using resolution attributes — such as layer thickness or dots per inch — as a careless shorthand term for accuracy.
Resolution does not, in fact, translate directly to a system’s overall accuracy. One simple way to visualize the real-world distinction between the two concepts is to imagine that two measuring sticks of differing length are both marked as 12-in. long, yet the second stick is actually 1-in. shorter. The first stick is divided into 1/16-in. increments, and its true length is verified at precisely 12 in. Even though the shorter stick reads “12 in.,” it is verified to be only 11-in. long. But this shorter measuring stick is divided into 1/32-in. increments, which is twice the resolution of the 12-in. measuring stick.
The 12-in. ruler with 1/16-in. increments exhibits lower resolution but high ultimate accuracy. The 11-in. ruler with the finer increments does the opposite: It exhibits high resolution but low absolute accuracy.
The same goes for additive-manufacturing equipment. Some machines make parts like the first ruler with finely resolved features but lower overall dimensional accuracy. Other machines turn out parts with lower, but acceptable, resolution and excellent accuracy.
When additive manufacturing was in its infancy and used primarily for model making, the distinction between accuracy and resolution did not matter much. It was enough for an early “rapid-prototyping” machine to turn out models that had approximate dimensions at best, as long as the model met the cosmetic goals. Users just needed a model to paint and show to their marketing department.
Today, though, much more is required of additive-manufacturing systems. They routinely turn out functional prototypes, fixtures, or finished goods that must meet the same stringent accuracy and repeatability standards associated with traditional manufacturing methods such as machining, injection molding, and casting.
The confusion between accuracy and resolution is understandable given the incremental way that additive-manufacturing machines create parts from CAD models. Some systems build parts from fused layers of a thermoplastic material. Some machines build parts from layers of a photopolymer that have been cured by a light source. Others bind or sinter layers of powdered materials, ranging from starches to metals.
Regardless of the build method, it’s tempting to equate the height of the incremental layer and the width of the smallest feature with system accuracy. The finer the resolution — the myth goes — the more accurate the part.
In some circumstances, there is limited truth to that reasoning. “In some applications, a fine resolution is important. In others, it doesn’t matter,” says Sheku Kamara, director of the rapid-prototyping lab at the Milwaukee School of Engineering.
So when does resolution count? “As it relates to accuracy, resolution becomes critical only when part feature size becomes very small,” says Kamara. When the application requires microscale feature sizes or wall thicknesses, then resolution can dictate a system’s ability to accurately build the small features. For example, Kamara points out, if the feature size is 0.002 in. and the system’s best Z resolution is 0.005 in., then layer thickness can constrain part-feature accuracy.This circumstance tends to arise in applications such as jewelry patterns and microfluidics components — applications needing very fine detail, which benefit from high-resolution equipment. Today, though, applications with microscale features are a tiny fraction of all additive-manufacturing jobs.
Most additive-manufacturing machines are used to build parts that measure several inches or more across and have tolerance capabilities no tighter than several thousandths of an inch. At this scale, the size of the material deposits or the “spot size” of the UV light doesn’t control the overall accuracy of the part or the repeatability of the process.
High-end fused-deposition-modeling (FDM) systems can produce parts with layer resolutions down to 0.005 in., delivering parts that aren’t as smooth as injection-molded parts. However, they have an overall accuracy or tolerance of ±0.0035 in. or ±0.0015 in./in., whichever is greater. This falls easily within the average tolerance for an injection molding job, which is typically 0.005 in.
If the additive-manufacturing machine’s most-important requirement is to produce a part with a class-A surface finish or sharp edges, then a high-resolution system may be necessary. But smooth surface finishes or sharp edges do not equate to accurate parts. Consider the following image. Part one was built on a 3D printer with a veryhigh layer resolution of 0.0015 in. (0.0381 mm). Part two was built on an FDM 3D printer using a lower-resolution setting of 0.010 in. (0.25 mm). As the FEA images of the parts show, accuracy is a function of the system’s capability to control the motion of the material deposition or curing mechanism across the entire build envelope.
The growing need for accuracy
Additive manufacturing got its start as a way to produce mostly cosmetic models but the field has evolved into full-fledged manufacturing. Today, additive systems turn out not just models for show and functional prototypes for physical testing, but finished goods, too.
FDM systems, for example, are increasingly employed as a cost-effective way to make manufacturing jigs, fixtures, and other tools. FDM machines also make low-volume production parts that in the past would have been injection molded or machined. Overall dimensional accuracy is paramount for these manufacturing aids and finished goods to function properly.
As additive parts move into applications with more-challenging functional requirements, their accuracy tolerances are specified in the same manner as those of traditional manufacturing methods. These tolerances are cited in thousandths of an inch (or hundredths of a millimeter) over given part dimensions, not dots per inch or slice height.
However, in manufacturing applications, the accuracy of a single part as it comes out of the system is just one of three critical considerations. The others are the repeatability of that accuracy over many parts and the stability of part dimensions over time.
According to Kamara, repeatability can make or break an additive system in functional-prototyping and direct-manufacturing applications, where multiple versions of parts must be made within acceptable partto- part dimensional tolerances. “Just as resolution does not translate into accuracy, accuracy does not translate into repeatability. Some systems have good accuracy but poor repeatability,” he says.
Kamara cites three repeatability considerations, part to part in a single build on a single machine, part to part in multiple builds on a single machine, and part to part in multiple builds on multiple machines.
Consistency across the build envelope, from build to build, and machine to machine is critical when manufacturing finished goods. Without process control, dimensional variance will yield unacceptable parts.
To scrutinize the repeatability of the FDM process, two studies analyzed thousands of dimensions over hundreds of parts manufactured on multiple systems. One study showed that the large-format, production- oriented FDM machine had a standard deviation of just 0.0017 in., which means that 99.5% of all dimensions were within ±0.005in. The other study showed that the multipurpose — prototyping and production — systems studied produced 95.4% of all dimensions within ±0.005 in., for a standard deviation of 0.0027 in. The high repeatability of these FDM systems is paired with long-term dimensional stability.
Whether a part has a service life of weeks or years, the repeatable accuracy of an additive-manufacturing machine is only half the equation. Just as critical is the material stability, which is responsible for part accuracy over time. Environmental conditions, such as heat, moisture, and UV exposure, as well as residual stresses from some additive processes may cause parts to shrink, expand, or warp.
Some processes use materials, such as photopolymers, that are less dimensionally stable over time than thermoplastics. “The materials experience changes in dimensions and mechanical properties, even after the part has been removed from the machine,” explains Kamara. However, additive parts made from industrial thermoplastics, such as ABS or polycarbonate, do not exhibit these postbuild changes.
The best way to assess the ultimate accuracy of an additive-manufacturing process is to measure parts over time. If you need parts that maintain their tolerance for months or years, don’t accept the measurements of newly produced parts. Plan a series of checks over an appropriate period to verify the material is stable.
For example, three random parts from the previously mentioned FDM repeatability studies were recently reevaluated. The parts had been haphazardly stored for well over a year, with no concern for environmental conditions. Yet the samples were almost unchanged. There was no warping, and the dimensions fall within the range of the original study. The lengths are within ±0.002 in. of the 5.000-in. nominal dimension, as was true in the study. Likewise, the 3.000-in. widths are within the same range of –0.003 to 0.004 in.
And separate studies by Loughborough University proved that FDM’s thermoplastics are just as stable in terms of mechanical properties.