Just Print it!

The dictionary defines the verb disrupt as “breaking apart” or “throwing something into disorder,” while the adjective disruptive is now often used to describe products, services, or ideas that radically change how we live, work, interact, or make things: The wheel comes to mind, the bicycle, steam power, the telephone, the car, aviation, penicillin, computers, and the Internet. Boatbuilders, no doubt, would add fiberglass to that list. Regardless, the promised benefits of disruptive technologies always include some combination of easier, better, faster, and cheaper.

Stephen Wu

Two parts of the construction mold of a 34′ power catamaran are being printed at Oakridge National Lab.

If former Microsoft executive Stephen Wu, who previously led two venture capital funds for the Carlyle Group, got this bet right, production boatbuilding should brace for profound disruption. Wu, an engineer by training, knows how to conceptualize technology that creates, or adds, tangible value. In this case, he used Big Area Additive Manufacturing (BAAM)—a loose term for 3D-printing large objects like cars and houses—to manufacture the construction mold for a 34′ (10.4m) power catamaran. Granted, boat builders and designers have been printing small plastic and metal parts for some time now to speed up prototyping (also see Printing Small Parts). But compared to what Wu tackled, those are itty-bitty pieces.

Forging partnerships to reduce cost

It was only in the spring of 2016 that Hanseyachts in Germany announced it had printed the hull for a Hanse 315 from wood. It was a clever and effective April Fools’ joke that would fall totally flat today, because at the time of this writing, the first boat to come out of a 3D-printed mold was well under way at XPlora Yachts LLC, a spin-off boatbuilding operation founded by Wu and his partners. Some specifics about the vessel were protected by strategic nondisclosure agreements but will be discussed in these pages as they become available. We learned that the boat will be powered by two 200-hp diesel inboards and be equipped with an advanced electrical system and convenience features such as a WiFi-enabled HVAC system that communicates with digital assistants like Apple’s Siri or Amazon’s Alexa. For now, we’ll focus on how expenses are being reduced, and the pains associated with printed tooling.


Stephen Wu

In 2017 Wu and his incubator Alliance Management Group LLC (AMG) entered a user agreement with the U.S. Department of Energy (DOE) Oak Ridge National Laboratory (ORNL) in Tennessee to prove that additive manufacturing technology is not just up to the job of printing tooling but is also faster, cheaper, and less wasteful than traditional subtractive methods that employ a 5-axis router to machine a large piece of foam into a plug, from which a fiberglass female mold is built. As a side benefit, Wu also wanted to eliminate or at least drastically reduce extensive, and expensive, fairing and coating. “Our goal was to make production boats more affordable, which means changing the most cost- and labor-intensive step—the building of the mold,” Wu told me in Kirkland, Washington. “Traditionally, that task used to take several months and cost hundreds of thousands of dollars for a boat of our type and size.” Simplifying this process, he explained, creates a win-win situation for industry and consumers. By yielding quantifiable savings in time and money for producers (he estimated across-the-board benefits of 30% to 80%), that could help boost builders’ profits while lowering the cost of the finished product, thus making boating more affordable.

Learning from Boeing

Visionaries must focus on the upside of disruptive technologies, which tends to hit snags when bold ideas move from the page (or the computer screen) to the shop floor. Elon Musk’s uphill struggle with the production of Tesla’s Model 3 is a recent instructive example. Wu’s approach, however, was inspired by his time at Microsoft back in the 1990s, when he led the Microsoft Consulting Services project team that helped Boeing develop the cabin-management system on its then-new Boeing 777 aircraft. “Boeing is less an aircraft builder than an aircraft innovator,” he said. “They don’t build the entire plane themselves, but focus on the most critical parts, like the wings.” In the process he learned strict manufacturing discipline and large-scale cooperation with hundreds of subcontractors.

Following Boeing’s model, Wu forged several partnerships for this project and entered a user agreement with ORNL, which gave him access to the Manufacturing Demonstration Facility (MDF) in Knoxville, Tennessee, to explore 3D printing for large tooling and allowed him to share the lessons learned in the process. The DOE, he said, is interested in energy savings and waste reduction, benefits that can be achieved through additive manufacturing.

“Tooling, tooling, tooling—and rapid prototyping,” quipped Dr. Brian Post answering my question about the best BAAM applications. He’s a research staff member with the Manufacturing Systems Research Group at ORNL who worked with Wu on the mold-printing project. “Additive manufacturing excels at high complexity and low volume; therefore, 3D printing can significantly reduce time and cost in the production of tooling,” he said. To showcase its capabilities to the auto industry, ORNL also printed a replica of the famous Shelby Cobra sports car, while pushing material development that helped significantly reduce the cost of printing material, from $50/lb–$200/lb initially to $1/lb to $5/lb, by switching the feedstock from filament to pellets, Post explained.


Dr. Brian Post

Printing Parts

With Wu’s surface data, ORNL set to work to print the 34′ mold. However, the Cincinnati BAAM printer, then one of the largest 3D printers in the world, was limited to a print envelope 8′ wide, 20′ long, and 6′ tall (2.44m x 6.10m x 1.83m), which meant that the printing had to be completed in six discrete 6′ sections. To achieve the best fit between the printed part and the computer model supplied by XPlora Yachts, the mold was printed vertically. The fine edges of the stern, which limited the access for machining the surface, meant that each section had to be split into two halves, for a total of 12 parts. Because the print head has a limited range of movement on the z-axis, the part itself had to be moved up and down as well. ORNL could print three mold parts simultaneously in approximately 12 hours. Calculating machine time, the mold sections were printed over a five-day period with 5,500 lbs (2,495 kg) of material, which cost approximately $5/lb, bringing material cost of the mold to $27,500.

Stephen Wu

Four monitors display the surveillance video (top row) and essential print parameters while the process is under way

It is critical to ensure a strong bond between the layers of printed material. The minimum layer-time is a “Goldilocks problem,” as Post called it. “If it cools too long, the bond between layers gets too weak. If it’s put down too fast, it’s still hot and gets crushed under the weight of the next layer. To control thermal distortion, we added carbon fiber, which has a very low coefficient of central thermal expansion [CTE] along its axis.” Fibers connect the layers, while resin acts as a bonding agent, Post explained. Adding carbon fiber improved tensile strength to between 8 ksi and 10 ksi (kilopounds per square inch) from 3 ksi with just polymer.

The printing material was acrylo­nitrile butadiene styrene (ABS), from Techmer PM (Electrafil J-1200/CF/20 3DP), which included 20% chopped carbon fiber in pellet form. A 0.3 (7.62mm) nozzle yielded a 0.15″ (3.81mm) layer height and 0.34″ (8.64mm) bead width. The printer’s gantry moved at 10.83 in/sec (27.50cm/sec) with a flow rate of 78 lbs/hr (35.38 kg/hr). In the end, the surface finish had a ribbed look and feel, but the printout was designed with an extra 0.15″ of material, which was later machined down to a smooth surface. “The cost of the machine was $150/hr, and we ran it in two shifts of 16 hours per day during the print phase,” Post said. “No operator was necessary, but for safety reasons we had one on duty.” Overall cost was divided roughly into 30% for material, 30% machine, and the rest for design and labor.

Some Subtraction

After printing, the project switched to subtractive methods: the mold parts were post-processed with a Thermwood router. A Faro laser tracker positioned and calibrated the printed part on the router table, and a tracking ball was moved over the mold surfaces to produce a so-called point cloud to be loaded into Verisurf, a measuring program that determined the best fit between the model and the printed part. The printed-model data were loaded into VisualMill, the tooling module of the VisualCAD/CAM software suite, before a 0.5″ (12.7mm) ball-end mill with 0.05″ (1.27mm) step-over paths traced the curved surfaces. The resulting tool paths were loaded into the router, which machined the mold to a smooth surface finish.


A five-axis router is machining the surface of a printed mold part

In some applications, it’s possible to skip machining by switching print resolutions from low to high, Post said. Smoothness depends on nozzle size and print speed. A large nozzle that extrudes material at a high speed, or deposition rate, produces a low resolution, with 100 lbs (45.36 kg) of material used per hour. By contrast, a small nozzle operating at slow speed depositing 20–30 lbs/hr (9.07–13.6 kg/hr)  results in a much finer print, i.e., higher resolution. But temperature is important, too. The critical temperature for the molding compound’s exothermic reaction is 230°F (110°C), Post said.

Making the Mold

With printing and machining complete, Wu had 12 smooth pieces to be assembled into a 34′ construction mold. First, three short rods were inserted horizontally to connect the port and starboard halves of each section. Then the resulting six sections were assembled and secured with threaded rods that run the full length of the mold. In addition, technicians also applied PlioGrip Plastic Repair 10 epoxy to the seams. That adhesive was chosen because it is specifically designed for ABS, with a 60-minute cure time, allowing alignment adjustments during assembly. Small inner tubes placed under the mold made it easier to slightly adjust alignment. To support the structure and maintain alignment, wood beams at the top of the mold controlled the span across, with tensioning straps holding the two sections together while the epoxy cured. Assembly of each section took approximately three hours, and the Plio­Grip fully cured in 24 hours.


Workers are applying epoxy to join the surfaces of two mold parts

The final step was assembling all sections to make the full mold, a process that started at the stern and worked forward to the bow sections. To help distribute the 10,000-lb (4,536-kg) load of the long tensioning rods, the crew manufactured aluminum bulkheads and screwed them onto the end sections. Each threaded rod protruding through the bulkheads was retained by a spring, a washer, and a nut. The springs enabled controlled tensioning on the rods to ensure even load distribution. To understand the design steps for the mold, Post suggested the example of concrete bridges: They have poor tensile-strength properties but are strong in compression, so cables are run through the structure to pull it tight.

The last addition was a pair of steel beams with lifting eyes and attachable casters at the bottom of the mold to help load it onto a flatbed truck equipped with air suspension. Before the long journey to the prototyping facility in Texas, a last quality-assurance measure was to scan the mold with a Faro laser-tracking system and compare the data to the original CAD model. The numbers showed that the fit had an average deviation of less than 0.050″/1.27mm.


The full project crew at ORNL with the finished product. Not all of them were on the job all the time, however.

A crew of three completed the mold printing in an elapsed time of two weeks of eight-hour days, and assembling the pieces (with up to 10 crew) took two to three days before everything was precisely aligned and securely joined.

Infusing the Hull

The composites work was done by the Composites Consulting Group, in Desoto, Texas. CCG has offices in eight countries, including Australia, China, India, Spain, Sweden, and the U.S., and works with mechanical and process engineers, material scientists, naval architects, and composites technicians. The firm designs and engineers details for composite projects that are efficient to implement on the factory floor.

“This project was the first with a 3D-printed mold,” said CCG’s regional technical manager, Belle Blanding. While she was not aware of any change in actual usage, she noted that “the outside of a printed mold is different, with a nicer geometry, and more accessible for the lack of the usual external structures on conventional molds.”

Stephen Wu

The first hull in the printed mold is getting infused with epoxy resin at the facilities of CCG

Before the 3D-printed mold could be used for infusion, its surfaces were brought to a 1,500-grit finish by air pressure, abrasives, and a lot of buffing. Next, the inside was coated with Duratec tooling gel and Loctite Frekote 770 mold-release agent. Then one small hull section was infused as a test, and those results were applied when fabrics and core were laid into the mold for the entire hull. CCG employed biaxial E-glass and Kevlar over high-density closed-cell Diviny­cell HM, and—because it has extra resistance against high engineroom temperatures—Divinycell HP marine-grade foam core, and high-temperature epoxy by Swiss manufacturer Hexion. Infusion time was one hour, with the vacuum sustained overnight to allow the resin to cure at an ambient temperature of 72°F (22.22°C). The prototype hull construction in the 3D-printed mold took eight days, Blanding said, “including the longitudinal stringers that added a couple of days.”

Bumps in the Road

Now here’s something skeptics can hang their hat on: Disruption is not a linear process. Searching for the leading edge, disruptors can easily find themselves on the bleeding edge as well. It’s part of the game, and this project had some of that, too. Here are three incidents that tested the structural integrity of the mold as well as the resolve and mental resilience of the stakeholders.

  • The project went well, but close inspection after de-molding revealed some pneumatic cracks that showed evidence of air leaks in two areas. The cause of the cracks, according to Stephen Wu, was the faulty application of the coating in the mold with a high-volume, low-pressure sprayer, so the cracks transferred to the hull. “However, these are hairline cracks and do not impact the integrity of the mold,” the report stated. “This issue can be solved by either reapplying the coating to the hull after every part is pulled or by applying a stronger or slightly thicker coating.” Blanding considered a 3D-printed mold “a step forward and upward [that] may be strong enough for production purposes,” but also thought that vacuum integrity would be better if the mold’s sections were fused to each other individually with a traditional bolt-and-flange assembly rather than by two enormously long tensioning rods.
  • Transporting this unique mold first to a composites show and then across the country to Washington State, where the initial production took place, proved Blanding’s concerns well founded, albeit in unforeseen ways. First, a hoisting strap broke, dropping the mold 18 (45.7cm) and cracking it in the bow area. It was repaired without much fuss
Stephen Wu

A failing hoisting strap caused the mold to drop 18 inches, sustaining a crack in the area behind the bow. The damage was repaired.

  • Another problem was traced back to forceful tightening of the tie-down straps to secure the mold to the truck bed during the 2,200-mile (3,541-km) journey from Texas to Washington. At the end of the trip the epoxied seams between two mold sections farther aft had opened up. An ORNL-crew flew in to de-tension the longitudinal tightening rods and realign the tooling. “It was fortunate,” Wu reported, “as soon as the rods were undone, [the pieces] wanted to fall back into line.”

Short Cycles of Innovation

Even as Wu’s print project was under way, ORNL and Ingersoll Machine Tools Inc. (Rockford, Illinois) were collaborating on a so-called Wide and High Additive Manufacturing (WHAM) system that can lay down as much as 1,000 lbs/hr (453 kg/hr) of print material on a build envelope of 20′wide x 8′ high x 60′ long (6.1m x 2.44m x 18.29m). This system, large enough to print tooling for airplane parts, can churn out Wu’s mold in two parts that need only to be joined at the keel like the shells of a clam. The new machine also includes an automatic exchange of the printing extruder with a high-speed 5-axis milling attachment for the conventional (subtractive) finishing step.

Thousands of materials can be used for 3D printing, including concrete, metals such as titanium, and, of course, polymers. The WHAM printer can use an amorphous polymer reinforced with 10% chopped carbon fiber, similar to what ORNL used for the XPlora mold. But the next step is already on the horizon. By adding recycled carbon fiber to the feedstock—a development funded by the DOE as part of a waste-reduction program—Post expects the price of low-modulus carbon to lower from around $10/lb (0.453 kg) to $5/lb. That would further accelerate the use of this material for marine applications and other leisure activities such as bicycling.

No Printer, No Problem

Even as the hardware becomes better and more affordable, there is room for a 3D-printing service—a high-tech equivalent of your neighborhood Kinkos. Additive Engineering Solutions (AES) in Akron, Ohio, claims to be the world’s first contract manufacturer to offer large-scale 3D-printing services for the aerospace, automotive, energy, and marine sectors. “To date, we have not built any molds for boats, but we have built molds for fiberglass and carbon fiber applications,” said Andrew Bader, vice president and co-founder of AES. Applications include tools and molds (layup and trim tooling for composites, autoclave tooling, stretch and form tooling), large-scale mock-ups, and prototypes. AES also offers a variety of specific services, from designing and (re)engineering parts to optimize them for 3D-printing, i.e., with full parametric 3D-modeling and computer-aided design (CAD) to materials development, precision machining, and custom coatings, i.e., for carbon fiber or fiberglass layup tools, including autoclave. At a 12′ x 5.5′ x 6′ (5.44m x 2.49m x 1.83m) build volume, AES can print materials up to 80 lbs/hr (36.29 kg/hr) using different nozzle sizes for different print resolutions. Their 5-axis router with a 10′ x 20′  (3.05m x 6.1m) bed can accommodate large parts and tools and can take near-net prints and machine them to a tolerance of 0.005″ (0.127mm).

  • 3D printing mold parts at ORNL Stephen Wu

The rate of progress reflects that 3D printing is still barely out of its infancy. As technology and materials evolve and the learning curve begins to flatten, processes will be optimized, accepted, and broadly applied, because in the end, the promise of faster, better, and cheaper is simply too attractive to pass up. Look for 3D printing’s impact on refit projects that could save chunks of money by printing replacement hardware (the entire part, not just the molds) from a digital file or from a 3D scan. Imagine how this technology could improve profit margins, shorten model cycles, and boost innovation and consumer interest.

Thus far, changes associated with 3D printing are mostly evident in the high-end niche, but decreasing costs and increased usability mean it will have a place in the builder’s toolbox, especially when the technology moves from prototyping and tooling to production of finished components. Getting there won’t happen overnight, nor will it be painless, but the industry as a whole will ride this innovation into the future, just as it did six decades ago when building vessels from molds by using rolls of synthetic cloth and buckets of resin became the new thing. Then as now, change was driven by the desire to make boats more profitable to build and more affordable to buy. Hard to find fault with that.

Lessons Learned

Despite a hiccup in demolding, this project demonstrated that a freestanding mold for a 34′ (10.4m) power catamaran with symmetrical hulls can be printed without adding thick coatings or a layer of fiberglass to the mold to produce fair parts, if these critical steps are studiously observed:

  • “Printers usually are not composite experts, and builders don’t have 3D-printing expertise,” the mold’s builder, Stephen Wu, noted. “Therefore, it is important that they continuously interact and exchange information to ensure the desired outcome.”
  • Designing a 3D-printed part is different. You can’t think like a boatbuilder but must envision the part as it is being printed. This is an additive process that requires contemplation of process and structure.
  • The design must support the structure of the printed mold and the part being made.
  • The mold must have a minimum number of sections if the 3D printer cannot print the entire mold at once. Wu says the next generation of large 3D-printing systems could print this size mold in two, not 12 parts. And Belle Blanding, of the Composites Consulting Group, which did the composites work, adds that fewer parts also would make the mold easier to seal for vacuum.
  • Bracing and tensioning the mold are critical, as it will come under significant mechanical stress during transport or in use.
  • Don’t drop the mold.
  • The nature of 3D-printing provides only a mechanical bond between printed layers, which may not be airtight. If infusion is to be used, proper sealing of the mold surface is paramount.
  • If the mold will be exposed to heat or chemical exotherm, processing temperatures should not exceed the thermal capabilities of the printing material.
  • If molds with complex curves are printed, demolding could be problematic.
  • Estimated life cycle is 10–15 years, but the structure would degrade under UV light if stored outdoors. The extent of mold maintenance has yet to be established and is the subject of future analysis. Wu said he sees no limit to the number of parts that can be pulled from that mold as long as it is properly stored (i.e., out of direct sunlight) and cared for. 
  • Some questions that only time will be able to answer include: What are the size limitations? What is the real-life expectancy for a 3D-printed mold, i.e., how will the shape hold up? How do builders interact with 3D printers to ensure the desired outcome? Will current mold makers adopt the technology and add it to their repertoire? How many outfits will have the equipment to provide this service to smaller builders who cannot afford to operate their own 3D printer?

The end.
About the Author: Dieter Loibner is an editor-at-large of Professional BoatBuilder.

For more on the advances in 3D printing in boat building and design, read the two companion pieces Printing Small Parts, and Print Yourself a Boat that previews the attempts of two Italian inventors to 3D-print a Mini 650 racing yacht that is scheduled to participate in the Mini-Transat race in 2019.