Skip to content

Time for Monumental Thinking in Additive

By Karen Haywood Queen Contributing Editor, SME Media

The limits of CAD and CAE tools and the STL file format are holding manufacturers back

As additive manufacturing emerges from a long infancy, the industry is grappling with a key challenge: A file format and design tools from the 20th century are being asked to do 21st century jobs.

“The industry was a hobby industry for 25 years and it’s starting to grow up,” said Kirk Rogers, technology leader at GE.

“You made a 3D model and it had a cool factor—Mickey Mouse or a little chess piece—it was awesome to look at,” said Kenneth Church, CEO of the R&D engineering firm Sciperio, as well as the 3D printing firm nScrypt. “What if I made it functional? Then the niche was rapid prototyping—get something out there and see how it fits. Then we moved from rapid prototyping to small-lot manufacturing.”

SM-16-Fall-Cover-2.jpg

But to fulfill what industry leaders say is its destiny, the technology involved must mature.

“We’re using old technology,” Church said. “From a hardware and mechanical point of view, it’s a very mature, stable technology in how we extrude, dispense and laser process. But it’s not so mature from a functional side that moves beyond mechanics and structure. When you start talking quantity, 3D printing is not the answer.”

For centuries, dreamers have designed things that could not yet be built. Think about Leonardo DaVinci’s helicopter on paper in the late 1400s, more than 400 years before the first successful flight.

Three decades after Chuck Hull invented the 3D printer and founded 3D Systems, that paradigm has shifted—at least in additive manufacturing (AM).

“For most of human history, we could design crazy shapes but there was not a way to make them,” said Hod Lipson, professor of mechanical engineering at Columbia University and chair of the AM data interchange committee for ASTM, one of the largest international standards development organizations. “For the first time in history, manufacturing is more advanced than design. With additive manufacturing, you can make almost any shape you can imagine. The challenge is the software design tools are not keeping pace. We can build anything but we can’t design it.”

Today’s high-end 3D printers have the capability to be more precise than the software and format used to design printable objects.

AM has been in dire need of audacious and monumental thinking—the sort that would land problem solvers’ visages on a 3D-printed Mount Rushmore. Fortunately, key people involved in multiple aspects of the industry are responding.

“If you get down to two microns and you find the error is in your STL model, then in a sense you’re defeating yourself by making the machine tools better,” said Jim Williams, chairman of ANSI’s American Makes Standards Collaborative (AMSC). He is also the former president of Paramount Industries and former VP for AM in aerospace and defense at 3D Systems.

On the front end, AM relies on CAD software, invented in 1961. That software is paired with the STL file format, invented in the mid-1980s to enable CAD software to transmit files to print 3D objects.

AM has evolved from its gee-whiz stage to an industry that aspires to move to mass production. It also seeks to be involved with multi-materials and biological structures, food, electrical functions and textiles. Another goal would be to play a larger role in other manufacturing sectors like automotive, aerospace and medical devices. However, the limits of CAD and the STL file format have become clear, industry leaders say.

“STL is just triangle geometry—nothing else,” said Justin Kidder, director of software architecture at Dassault Systemes.

“STL is a very simple format designed to work on a computer of almost 30 years ago,” Rogers said.
“It has not transitioned well into those other genres,” Church said.

As for widespread use of a file format to replace STL, Met-L-Flo President Carl Dekker predicted as long as 15 years. “I don’t want to believe that answer,” he said. “But historically, seeing how long it takes things to be adopted and become integrated, I fear it’s going to be a decade or maybe a decade and a half. I hope it’s going to be a lot faster than that.”

“An ideal timeline is yesterday,” Kidder said. “But realistically it’s going to take time for adoption to grow in the market.”

Simple was good, for a time

The STL format did solve the major issue in the 1980s—simplifying CAD data, Dekker said.

Circuit_Board-for-SM16-Fall-768x808.jpg
Today, it would take any manufacturer longer than a month to 3D print an electronic phased array antenna (PAA) element. While a CAD program would be involved, patterns and commands would need to be entered by hand, nScrypt CEO Kenneth Church said. If the industry had better software, the RF multi-layer circuit, 7 layer electronic board could be printed in hours. PAAs electronically steer the antenna without physically moving it. On the drawing board with the DOD and the University of South Florida: Printing doubly curved phased arrays, which allow the beam to steer past 180 degrees.

“These systems would take hours to process files that we don’t even think twice now about sending as an email attachment,” he said. “At that point, it was an amazing ability to be able to manufacture a part through the process. We’re still using the same data set. They made it as simple as possible and no one has changed it since.”

Using STL, a design in CAD is exported as an STL file, which describes a three-dimensional object as a series of linked triangles. The data has to go through at least one additional step to slicer software that converts the digital 3D models into printing instructions.

“Circles can’t be cut nicely into one or two triangles,” Rogers said. “It takes 20 or 30 triangles or more to get the right resolution of that circular shape, which causes file sizes to blow up huge when trying to represent complex geometry to something the printer would understand.”

“As I was coming up to speed on 3D printing, it was obvious there were serious impediments for success for the industry,” said Adrian Lannin, Microsoft group program manager who worked first with paper printing and has worked with 3D printing since 2013. “I saw there were things people were having to do to make 3D printing work that weren’t productive. STL is not efficient for 3D printing. It’s error prone. We’re looking at places where we can improve the productivity of additive manufacturing, make it less of an IT headache by having the pieces naturally work together.”

Limitations of the STL file format include:

  • Accurately defining complex, complicated or large geometric shapes and structures can be difficult and involves creating large files, 10-20 GB of data. These files take a long time to move from the designer’s desk to a 3D printer, which may not even be able to accept the entire file.
  • The format does not specify units—inches, centimeters, millimeters—being used. Those specs have to be sent separately, which opens up the possibility of errors.
  • The format does not specify colors, textures or material.
  • In fact, the STL format cannot embed any other data beyond the design, including information related to copyright and file security.
  • Modifying the file is difficult. The file format can’t distinguish between minor and major changes so any change means the entire work flow must start over, which can add hours to the design process.
  • Designs have to pass through a number of steps and translations from concept to end result, which again creates the possibility for errors.

“There are a lot of things those files don’t contain,” said Shaun Kroeger, director of partner sales for the Americas at software company solidThinking.

“We need a file format that can hold all the information—not just the overall shape,” said Mark Rushton, product portfolio manager at Dassault Systemes.

“The challenge is exporting data from your design software that is accurate enough, while maintaining a file size that is small enough to recreate finite part details. Data can be lost in translation,” said Michael Hansen, applications engineer at SLM Solutions, which works in selective laser melting and other AM technology. “The data may not be exact enough or the exported data may contain errors.”

Two replacements emerging

Key players in the industry have come together to develop replacements for STL: AMF and 3MF.

Each format makes it easier to represent other geometric shapes, such as arcs, Dekker said.

Unlike STL, 3MF and AMF files also represent material, texture, colors, authorship and other critical data concisely and unambiguously within the file itself, Kidder said.

“The data lives with the file,” he said.

AMF work began in 2010

AMF is focused on a comprehensive, standards-driven format.

AMF (aka ISO/ASTM 52915) was developed starting in 2010 through ASTM. It was first approved in 2011 and has undergone several revisions, said Pat Picariello, director of development operations for ASTM International, a not-for-profit forum that provides a consensus for the development of standards. Also involved is America Makes, which the Obama administration established in 2012.

More AM file format standards are needed, said Williams, also former chairman of America Makes’ executive committee. Only a handful of standards specific to AM existed at the start of this year, and of those very few 3D printing file format standards exist.

For example, still needed are standards associated with operator training, machine calibration, machine validation and final part inspection, Dekker said.

But standards should be written carefully so as not to restrict the evolution of the format, he said. “You don’t want to write standards to create handcuffs. There’s no way to tell whether in five years, someone may come out with a way that is absolutely the best way to do it. But have you eliminated the ability for that to come forward because the standard was written in such a restraining way?”

Users like AMF. But adoption has not been widespread.

“Feedback from some potential users was that AMF is great, but it’s almost too sophisticated and too broad to adopt in a change perspective in a relatively short period of time,” Picariello said. “It scared some people off.”

Meantime, 3MF is “more focused, more scaled down, more readily accessible” and yet “not meant to replace AMF,” Picariello said.

3MF work began in 2011

3MF was inspired by AMF and is focused on quickly getting a simple but extensible format in widespread use and moving toward standards later.

Microsoft began working on 3MF in 2011 with the goal of creating a format to work in Windows that was easy to use, unambiguous and complete in itself, Lannin said. Microsoft originally offered the file format as licensed software that people had to pay to use, but the cost was a stumbling block to wider adoption. In 2014, Microsoft began offering the software as open source.

The 3MF consortium worked on its specs for about six months, first published initial file specs in 2015 and is focused on getting 3MF in play this year and next, Lannin said.

Software firms are taking note.

3MF Consortium member companies are implementing extensions and adding capabilities to their software packages, which then may be published by the consortium, Kidder and Lannin said.

One 3MF extension in particular, for 2D slice data, means that instead of using slicer software to represent the object in the final steps before an object is printed, the slicing could be done earlier in the work flow, Kidder said. That eliminates a step, reduces the probability of error and defines slicer data within the file itself, he said.

As for standards, the 3MF Consortium recognizes that at some point, 3MF should become an official standard, Lannin said. “We have worked with ASTM to discuss what kind of standardization we want to go with in the future. For now, we want to be a bit more nimble.”

‘Coopetition’ seen on horizon

Developers of both file formats also are working to foster collaboration.

ASTM in 2013 agreed to develop standards needed for projects funded by America Makes to move forward, Picariello said. The 3MF Consortium and ASTM also agreed to enable a pipeline for ideas between the two formats, he said.

Major software firms are working to ensure their offerings integrate with at least one or both of the formats. For example, the next version of SolidWorks will support 3MF, Rushton said.

Software another big issue

Software challenges are also holding AM back.

Current CAD system software can’t keep up with all the materials that could be used in AM to, for example, make a material like human bone, make a bird’s beak or make a prototype of a baby’s heart for a surgeon to practice on, Lipson said.

“Computer-aided engineering [CAE] analysis tools are not keeping up with the capabilities of additive manufacturing,” he said. “If they don’t change, new companies will come around with new software and eat their lunch.”

America Makes and the more recently established Flexible Hybrid Electronics Manufacturing Innovation Institute, aka NextFlex, recently put out calls for better software to support electronic 3D printing.

“They are starting to ask the intelligent questions, such as ‘Who is going to bring this software together?’,” Church said. “We have a few companies we are working with right now to get them to talk to each other, or at a minimum talk to each other through us.”

Several companies, including nScrypt, Autodesk, SolidWorks, CDS, Boeing and Raytheon, appear to be willing to come to the table to engage in “coopetition,” he said.

For example, Autodesk is offering more optimization tools within its software, such as with Fusion 360, and for simulation with the upcoming release of Netfabb 2017, said Duann Scott, business development manager, Autodesk Digital Manufacturing Group.

“We have been collaborating and competing with a number of companies in this business,” Church said. “We need each other to help push this industry forward. We are optimistic and this is not bad. What is bad is to over-promise and under-deliver. This hurts the industry. We must show that we can make electronically functional structures that bring value.”

Tackling parametric modeling, validation

“One major challenge is getting from an optimized shape to a smoothed out part without a lot of manual interaction,” solidThinking’s Kroeger said. “A lot of CAD programs have very little ability to edit tessellated geometry [repeating a shape with no overlaps or gaps].”

Early this year, solidThinking added polynurbs capability to its Inspire platform that enables users to modify a design file to transition directly to solid geometry from tessellated geometry, he said. This is critical, he said, because it is very tedious and time consuming—if not impossible—to model optimized organic surfaces using parametric modeling techniques.

Simulating the printed object earlier in the design process will help advance the industry.

With most current software, the object to be printed isn’t simulated until the final steps, Kroeger said.

“I have always preached design validation,” he said. “But the earlier in the design process a company can use simulation, the more value they can get from it.”

“We can see parallels with the adoption of digital photography,” said Ulf Lindhe, business development lead for AM at Autodesk. “With traditional analog photography, there was typically a one-day to one-week delay between when you took the photo and when you saw the results. With that, understanding aperture, film speed, composition, depth of field took trial and error, with a significant delay between the act and the feedback. This is where we are today with design for manufacturing.”

“Once digital photography became widely adopted, people would take a photo, see it immediately on their screen, adjust settings, take another shot and quickly learn how to take better photographs,” he said. “Now the gap between amateur and professional photography is very small. This is how we’re now approaching design for additive manufacturing, with simulation and optimization as an area of focus to help with enabling faster iteration and incorporating the exact machine and material parameters into the design file.”

Print validation tools are essential to improve the process, Rushton said.

“Will this part fit on my printer? Have I got features too small to actually print? Those are the kind of checks we need before we send it to print,” he said. “But every AM system works in different ways, and each machine has its own variation on materials. That makes it difficult to have a static set of checks to cover all technologies. This variation amplifies the complexity when it comes to simulating the physical performance of a design before you print it.”

Even on the same machine, “You can build one part in the middle of the build chamber and it will experience different residual stresses compared with a part built in a different place,” Rushton said.

“The software has to be very complicated to know all the outputs. We have the capability within the Dassault Systemes portfolio. We can do the calculations. But it’s knowing what figures to put in to start with, which is the difficult part. So many parameters can affect the final part’s performance and they all need to be known so the entire build process can be simulated in order to simulate the performance of the built part.”

Achieving these milestones will help shorten the design process.

“To better create a design optimized for a specific machine and material combination, we need to first understand what the material properties will be after it’s manufactured,” Autodesk’s Scott said. “To achieve this, we need in-process monitoring of the machine so that we can simulate the process and optimize not only the geometry but also the machine control. Once we have this real-time connectivity to the machine, we will be able to produce predictable, reliable and repeatable parts with significantly reduced trial and error. We have the optimization and simulation software in place and just need to work in feedback from the machines to help reach the potential of additive manufacturing.”

Process automation and optimization can be a way to achieve the goal of simplifying the AM simulation process and minimizing the gap between the designed and manufactured part, Rushton said.

“A simulation expert would develop, validate and simplify the simulation process, exposing only what a non-engineer would need as inputs and outputs,” he added.

A streamlined, easier-to-use process will enable non-engineers or others without a deep level of expertise to work in 3D printing, Scott and Church said.

The Inspire software has shifted simulation into early phases of design process, Kroeger said. Although the new technology costs more, the shorter design process outweighs the added cost, he said.

“Just using this technology, we’re seeing 200-300 percent improvement in the design throughput,” he said. “When it gets to that ideal situation, we’ll see another 200-300 percent off that. Once you get the shape and press a single button and go to the printer, you’ve eliminated parametric CAD and a good bit of the design process. Scientists at our company say we are 95 percent of the way to button-click-and-smooth use.”

Alpha software in a year?

Church predicted a year for alpha software. Kroeger predicted about five years. Once that happens, AM will be able to compete more effectively.

Compared with traditional manufacturing, AM offers more economical topology optimization, SLM Solutions’ Hansen said. Manufacturers using AM are not limited to producing round holes because of cylindrical tooling.

Similarly, with AM, manufacturers may be able to build parts using a lattice structure to realize the same or better strength, rigidity and fatigue life with 40-60 percent less material than what is used with traditionally manufactured parts, he said.

“In an ideal world, if I could take an engine block or suspension component for an automobile and remove excess material required for traditional manufacturing methods and still have the same strength and same mechanical properties, I’m saving money producing that part,” Hansen said. “Often, I can build it faster than a traditional part. Things wouldn’t necessarily need to be over-engineered.”

This article was first published in the Fall 2016 edition of Smart Manufacturing magazine.

  • VIEW ALL ARTICLES
  • Connect With Us
    TwitterFacebookLinkedInYouTube

Latest News

Always Stay Informed

Receive the latest manufacturing news and technical information by subscribing to our monthly and quarterly magazines, weekly and monthly eNewsletters, and podcast channel.