4.2 Design Optimization: Implementation while considering build constraints

Internal Passageways

When designing for AM processes such as PBF, binder jet, or polymer SLS, one of the major benefits of AM is the ability to create internal passageways, or open cavities that pass through the inside of the part. However, there are some nuances when designing these types of features. For example, AM PBF creates walls that are inherently rough. If the internal passageway needs to be smooth, expect to add cost with a post-processing step to smooth out the passageways. And even with the extra post-processing costs, don’t expect mirror finish smooth surfaces like those of machined parts.

Also for processes that require support material, the profile of an internal circular channel will most likely need to be MfAM’d so that it is self-supporting during the fabrication process. If support structures are left in the passageways, they may be difficult (or impossible) to remove and will impede the flow through the internal passageway. Note that redesigning internal passageways is not as much of an issue with polymer SLS and binder jet, where the powder alone is supportive enough without excessive solid support structures.

Complex Sweeps and Surfaces

With AM, geometries don’t need to be made using simple extrude, sweep, and revolve commands found in traditional 3D CAD software methods. Complex curves and surfaces can be used to generate unique geometries and organic shapes; provided designers know how to use the tools to create these structures.

It is important to maintain the orientation of the internal passageway, which is a diamond shape, so that it remains self-supporting as discussed in the previous section. Otherwise, the interior may need support structures which will be impossible to remove.

A Cellular feature, or type of mechanical structure that repeats a specific structural pattern, in order to offer a lightweight design solution. These repeating patterns encompass 3D or 2D space and are adjusted to accommodate different loading conditions. Of cellular features, two of the most common found in DfAM are honeycomb structures and lattice structures.

Honeycomb structures follow a form of biomimicry which draws inspiration from bee honeycombs. These types of structures are typically used in the aerospace industry made from metal or composite to form a core.

“Lattice structures are topologically ordered, three-dimensional open-celled structures composed of one or more repeating unit cells. These cells are defined by the dimensions and connectivity of their constituent strut elements, which are connected at specific nodes”. (L. Hao, et al., 2011). These types of structures fulfill specific stiffness requirements, and simultaneously achieve weight reduction relative to bulk structure. An AM example is shown in Figure 4.9.

Creating honeycomb or lattice is becoming a very popular AM design engineer’s tool in the toolbox as these types of features can be created in polymer SLS or Material Extrusion (FDM), as well as metal PBF and BJP technologies. However, there remains some challenges in using these structures in a part that exhibits high loading requirements or stringent inspection criteria.

Figure 4.9 An example of a lattice structure. (credit: Modification of “Form 2 closed lattice” by Creative Tools/Flickr CC BY 2.0)

First, these features are difficult to produce with traditional 3D CAD software. On the other hand, they can be produced, but should they be? Honeycombs, lattices, and other lightweight features are great for additive, but they can be difficult to create and analyze in current tools. Software such as nTopology makes the process of creating cellular features easier as they have approaches that generate the geometry in more efficient ways. The program easily computes and displays complex lattices and organic structures.

The second challenge is structural evaluation. Just as creating these structures is difficult in 3D CAD, they are even harder to mesh in FEA software. Simply exporting these types of structures into FEA software is a challenge. At a higher level, structural analysts see every surface profile as ‘defect’ potential for a stress fracture. The rougher the surface, the higher risk of failure, especially with high and low cycle fatigue conditions. With honeycomb, lattice or cellular features, the designer exponentially increases the potential of premature failure if any significant load is passed through these types of rough surface structures. On the other hand, as a way to increase heat transfer efficiency, that same increase in surface roughness might be a desirable feature for complex heat exchangers composed of lattice.

The third challenge is inspection. If a part consists of many cellular or lattice features, how will it be inspected for conformance to design intent in a cost-effective way? With the current state of inspection technology available, it would be difficult to find an inspection method that could be used on these types of structures in a serialized production environment.

From a design standpoint, if a designer were to use honeycomb, lattice or cellular features, the best approach would be to place them in non-load bearing areas of structures to avoid the critical evaluation and software challenges described above.

Isogrids

Isogrid features are stiffening ribs, typically triangular in shape, protruding from a surface at a defined distance from the surface. These features have been used for several years in industry as a light weighting practice for CNC machining and composite fabrication industries.

Figure 4.10 Isogrids on the pressure vessel for the CST-100 spacecraft. (credit: NASA, Public Domain)

Whereas lattice and cellular features are linked to PBF processes, isogriding offers flexibility across AM process types. This type of feature offers structural stiffness with reduced weight. It is also noteworthy that since applying isogrid features has been used in industry for a number of years, structural analysts will have greater ease analyzing this type of structure relative to cellular or lattice features. As a result, this type of design sees less certification scrutiny and analysis.

The discussed design features optimize AM shapes and performance, not without certain challenges. With those challenges and tips for overcoming them in mind, there is still more to master in design optimization as it pertains to the different AM fabrication processes. The remaining sections of this subject will outline basic design allowables according to each AM process as they exist today. Note that process development could always expand these guidelines in the future.

PBF Machine Process Constraints

When designing AM parts for PBF processes, there should be a focus on how the process may be constraining. This first obvious constraint is the overall build platform volume. This limits the size of AM part one can build in the process. Each year, equipment manufacturers develop machines with larger build volume sizes than the previous year. There can be large scaling differences for some AM processes. For example, an FDM type machine can exist in the range of 8.6 inches x 8.6 inches x 9.5 inches on the small end consumer printer to 240 inches x 90 inches x 72 inches for the large Cincinnati BAAM system. For PBF metals, one is the larger GE Concept Laser Xline with a build volume of 31.5 inches x 15.75 inches x 19.68 inches. As mentioned, these machine sizes will likely be surpassed by new models or newer processes each year. Most all AM equipment manufacturers will post the build volume sizes on their websites, so be sure to check the AM companies’ website on how large of an AM component can fit inside the machine to establish scale constraints.

DED Machine Process Constraints

Depending on the type of DED process, the scale of the machines may vary. Blown powder deposition machine sizes can beas large as 5 feet x 5 feet x 7 feet with wire fed machines as large as 19 feet in length, 4 feet wide, 4 feet high, and 8 feet in diameter for some machines. However, similar to the PBF, manufacturers are constantly increasing the physical size of the build volumes. AM equipment manufacturers will post the build volume sizes on their websites, so be sure to check the AM machine companies’ website on how large of an AM component can fit inside the DED machine to establish scale constraints for parts fabricated from AM DED.

Hybrid vs Conventional

Hybrid DED machines integrate a milling step during the layered process of fabricating parts and as a result, offer superior surface finished parts than conventional DED machines. This process approach allows for the AM part to exhibit machined finish across all surfaces of the part if desired. Traditional DED machines require a part to be milled after a part is produced and removed from the chamber.

The upside to using a hybrid DED process is the absence of a post-process milling operation and the avoidance of tooling setups for secondary milling operations. This allows for faster throughput of the overall AM fabrication cycle. However, the downside to this approach is the added complexity of programming requirements for a hybrid DED process and hybrid DED machines are typically smaller in build envelope compared to conventional DED machines due to the added complexities of the hybrid machine. The catch-all nature of the process also precludes the part from going through a stress relief cycle or heat treatment before machining, which can be problematic for final dimensions staying in tolerance.

Also, both hybrid and conventional blown powder DED machines offer the capability of a tilt table for the base plate, thereby allowing 5 axis manipulation of the baseplate. This is a significant advantage from a design standpoint. Recall that all DED machines produce geometry without support material. By fabricating a DED part on a machine with a tilt table, this allows for part features to be designed parallel to the baseplate thereby allowing the part to shift planes during fabrication. With this approach, designers will inherently have more design freedom and not have to design wasteful sacrificial support structure to stabilize part features during printing.

Wire Fed vs Blown Powder DED

DED systems have different deposition rates and use different material feedstocks (wire vs. powder), which have implications on design and post-processing. This allows for flexibility if a part need be constructed in freeform or have features added in a repair or modification scenario.

Typically, large scale systems deposit a thick melt puddle in scale and are wire fed. From a design perspective, if a part needs to have features less than .39 inches (10mm) thick, then recognize that these features would be difficult to produce using large scale wire DED systems without post-process milling operations. In addition, due to the relative thickness of the melt puddle, note that sharp transitioning, overhanging features are problematic to produce with wire fed DED systems.

However, there is an upside to large scale AM DED wire fed systems. For these larger metal components, DED may be the only answer. In Figure 4.11, nearly 400 lbs of material was deposited to make the 7-foot boom on this mini excavator. The boom was built vertically; so, the substrate was not integral to the structure. Holes were designed to be self-supporting, and the geometry was hollowed out and symmetric to mitigate warping during the build. After deposition, only the mating interfaces had to be machined at the connection points. “The 7-foot-long, 400 lbs. ‘stick’ was printed entirely of low-cost steel on the Wolf Robotics Wolf Pack printer in only 5 days.”

Features of this build included:

• “Throw Away” build substrate
• Hollow symmetrical structure mitigated distortion
• Holes designed with “tear drop” shape to accommodate self-supporting angle of DED process
• Minimal machining (i.e., mating surfaces)

Figure 4.11 Large-scale DED excavator boom 3D printed at the U.S. Department of Energy’s Manufacturing Demonstration Facility at Oak Ridge National Laboratory (credit: Modification of “Project AME” by Oak Ridge National Laboratory/Flickr CC BY 2.0)

For large scale wire fed DED design, it is very important to consider the build plate substrate as a sacrificial element to the design. This is why it is helpful to pick AM candidates that are symmetric about a build plate as the build plate becomes the part during the fabrication process.

Medium scale systems typically require blown powder feedstock and can be hybrid or utilize conventional DED processing. From a design approach, these blown powder DED machines allow for easy buildup and repair of conventionally manufactured parts. Compared to conventional repair technologies such as welding, blown powder DED yields exceptional metallurgical bonding, less warping and distortion, lower dilution rates, is capable of automation, and lower heat input due to a smaller heat affected zone.

However, the most fascinating aspect of blown powder DED systems is the ability to have multiple powder feedstock input streams. This allows for multiple materials to feed the laser deposition concurrently. In a practical sense, this allows designers to customize materials within a single part. Being able to gradually blend different composition of materials throughout a part produces what is known as Functionally Gradient Materials (FGM). From a design perspective imagine if a part had features that were 99% Tungsten to withstand abrasion or thermal loads, while on a different area of the same part could be comprised of 99% Titanium for its lightweight nature. Between the two areas of the part would be a gradual blend between Tungsten and Titanium. If these two materials were able to be functionally graded without adverse cracking effects, perhaps this could be possible. There exists an entire field of materials research looking into alloy and material compatibility for FGM. It is remarkable that a technology such as blown powder DED could enable the application of FGM into physical hardware.

The small-scale blown powder systems offer a finer deposition melt puddle width achieved through a blown powder, but are typically reserved for smaller laboratory scale research. These smaller systems offer most of the functional capability of the medium scale systems, however, in a smaller format machine size.

At the time of writing the below offers a snapshot of the manufacturers offering each type of system.

Manufacturers of wire fed large scale DED systems include:

• Sciaky
• Norsk Titanium
• WAAM
• Lincoln Electric
• EVOBEAM
• Prodways
• Gefertec

Manufacturers of blown powder large scale DED systems include:

• DMG Mori (hybrid)
• BeAM
• RPM Innovations
• Formalloy
• MAZAK (hybrid)

Manufacturers of blown powder small scale DED systems include:

• Optomec
• RPM Innovations
• Formalloy

Material Extrusion

Material Extrusion, commonly known as FDM, operates under very similar constraints as the other AM processes. While the general golden rule to build at 45 degrees or steeper still applies, design for this process can be more lenient as the support structures can be made to breakaway easily or in some cases, are liquid soluble. Much like other nozzle-driven processes, the smallest feature size is constrained to the nozzle diameter. No feature can be made smaller than the width of material laid. This makes complexity of the design, nozzle diameter dependent. This is important to keep in mind when designing for a specific machine. Another important acknowledgement of nozzle diameter is how it will affect the surface roughness. When building relatively small-scale parts, as Material Extrusion is often utilized for, the nozzle diameter can impact surface roughness greatly. If a rough surface is problematic to the application, it should be compensated for by inflating or adding stock to surfaces to have material removed later. Post-process material removal can be done in a variety of ways such as a chemical bath for uniform smoothing, CNC machining of specific surfaces, abrasive flow for passages, and so on. The exact process used to achieve surface smoothing goals should be planned for specifically. Much like it should be in any AM processes.