Why multi-material additive manufacturing will change our approach to part design

Share This Post

Share on facebook
Share on linkedin
Share on twitter
Share on email

If you’ve read our earlier Medium posts you’ll know that Aerosint is working to develop a selective powder deposition system with the goal of making additive manufacturing (AM) capable of building functional multi-material components. We’ve also covered some reasons why multi-material parts will advance the performance and simplify the assembly of complex, high-value devices and machinery. One important thing we haven’t yet discussed is how the emergence of multi-material AM will necessitate a big change in the computer-aided design (CAD) process that takes into account material allocation and characteristics. Below we’ll discuss the reasons why the CAD workflow is being rethought specifically for printing multi-material parts, and we’ll introduce you to several software companies already gearing up for the multi-material revolution.

Parametric CAD is showing its age

AM is accelerating in its technical maturity and adoption by industry. This growth is in large part due to the ever decreasing cost and improved capabilities of electronic hardware (stepper motors, lasers, CPUs, etc.) and control software used to automate and regulate these layer-wise material deposition processes. But while AM machine capabilities were steadily advancing, mainstream CAD software makers weren’t thinking much about how to alter their design paradigm to suit AM’s capabilities; the popular CAD suite SolidWorks didn’t release a topology optimization module until the 2017 release, nearly 30 years since the concept was introduced.

The typical CAD workflow is best suited to traditional manufacturing. Basic shapes and modification processes used to form parts in CAD reflect typical stock material geometries (plates, blocks, cylinders, tubes, etc.) and forming/machining methods (extrusion, cutting, filleting, chamfering). Also, because material stock pieces are uniform in composition throughout, solids in CAD are represented simply by defining their boundaries — the volume enclosed within one continuous boundary is assumed to be composed of one uniform material.

For single-material AM processes this workflow isn’t any particular issue (even if it doesn’t really take advantage of the geometric freedom AM offers). Multi-material printing, however, will require CAD software in which materials can be explicitly “allocated” at certain positions within the part volume to provide the desired functionality.

Allocating material within the part volume means that volumes can no longer be defined only by their boundaries. The alternative is to define volumes as a collection of smaller units of volume which can be individually designated as one material or another. These small volume units are called “voxels,” the 3-D analog of 2-D “pixels” that comprise digital images.

Who’s thinking ahead?

Several companies have recognized the need for voxel-based design and have developed “voxel engines” — software which enable design of parts consisting of voxel units and subsequent (or concurrent) designation of pixels as material A or material B (or C, or D, and so on, depending on the capabilities of the printing process). In many ways these software packages are ahead of their time since so few multi-material AM techniques exist, and those that do are primarily geared toward prototyping. By including advanced features like generative design (i.e. automated material allocation based on simple fixation and mechanical load requirements and material characteristics), it’s clear that these companies are looking toward a future in which multi-material AM will be used to produce advanced components for high-value industries and critical applications.

Who are these forward-looking folks and what have they developed? Below we briefly introduce several design software packages with explicit multi-material functionality or realistic possibility to include it.


To our knowledge Monolith was the first design software expressly developed for voxel-based design of multi-material components. It was developed by former Harvard Graduate School of Design researchers Panagiotis Michalatos and Andrew Payne. It seems to have been developed primarily for artistic and prototype-oriented use, especially in combination with the Objet / Stratasys material jetting systems that can print curable polymer resins in a multitude of colors and with varying material properties.

Monolith is rich in features that enable the user to create complex and beautiful designs. Voxelized geometries can be generated by free ‘painting’, selection and combination of predefined shapes, and sweeping 2D bitmaps along an axis and combining them. Functions can also be used to allocate voxels in space according to mathematical equations. Since each voxel contains information about its composition (i.e. either material A or material B), a two-material distribution can also be defined within the volume by free drawing, gradients, and a number of readymade functions and patterns.

Examples of the free voxel drawing function in Monolith (top), and actual parts printed on a Stratasys/Object Connex system (bottom), based on voxel-based models designed with Monolith.

For engineering applications where parts have been designed parametrically, there is an import STL option. This generates a thin shell which must be “filled’’ (i.e. designated to contain material voxels rather than empty space) before exotic things like gradients and patterns can be applied to the volume. Still, it provides a practical route for engineers to define part dimensions in a familiar way and perform material operations on the voxelized volume prior to multi-material printing.

The future of Monolith isn’t very clear. Even though it began as a mostly academic exercise, the Monolith project was acquired by CAD software giant Autodesk in 2015 — apparently because they saw some commercial potential in it — and Michalatos and Payne were hired by the company at the same time. Still, at the time of writing the software isn’t being sold as an official Autodesk product. You can actually download and try it out for free.


Like the Monolith project, nTopology was founded by an architect (Bradley Rothenberg). Unlike Monolith, nTopology’s software is quite explicitly targeted at engineers and designers. Their product, nTop Platform, is a software suite built with the advanced manufacturing of functional components in mind. nTop Platform can perform standalone design and analysis functions, but it can also import part geometry and mechanical simulation data from a variety of popular parametric CAD systems. CAD files are directly imported to nTop Platform rather than having to export them as STLs, as is the case for Monolith.

nTop Platform converts the parametric CAD files to combinations of “signed distance functions” — equations that return the shortest distance from a point to a surface — so that they can be easily manipulated by mathematical operation. Why would we want to represent 3D shapes using equations? For one, it allows shapes to be combined by “constructive solid geometry” using Boolean operations (AND, OR, NOT, AND NOT), in order to modify designs with lattice infill structures and other features that maintain part strength while reducing weight.

Screenshot of the nTop Platform software showing an impeller “field” (blue) modified with an gyroid infill pattern (red) to yield a more performant lightweight part which is only manufacturable using an AM method (image source: nTopology website)

Another reason is that since the shapes are defined by semi-continuous functions, it’s only when the models are exported for printing that the voxel resolution must be defined. This flexibility means that designers don’t need any a priori knowledge about an AM system’s resolution before creating a part, thus different versions with different voxel sizes can be exported for printing on different systems.

At this time the nTopology team is offering demos of the software to interested parties and is currently available for purchase. We were recently given a short demo by Andrew Sartorelli, an Application Engineer working with the Europe-based nTopology team. The interface is sleek and appears user-friendly, but there will undoubtedly be a learning curve to apply the more advanced features of the software, e.g. spatially varying the unit cell size of a weight-reducing lattice infill structure according to iterative stress analysis. The ability to assign different materials to different regions of the part according to a signed distance function is not yet part of the current nTop Platform build, but Sartorelli indicated that multi-material functionality will be included in a future version. Having recently raised a $20 Million investment round, New York-based nTopology seems well positioned to build out multi-material functionality in an upcoming release.


Hyperganic is a German startup company that seems to be halfway out of stealth mode. Founded in 2015, their goal is to drastically change the paradigm of component design to take full advantage of the geometric freedom of additive manufacturing by leaving much of the design process to their “voxel engine.”

The voxel engine is a generative algorithm that builds voxels where necessary in a volume to fulfill user-defined constraints like load conditions, fixation points, material density, etc. In this way, the design process is essentially flipped on its head — the algorithm, rather than the designer, decides where and how material is allocated. According to Hyperganic, Nature serves as their inspiration in how it evolves complex forms to achieve optimal functionality. Because AM makes it possible to manufacture functionally and topologically optimized components, highly efficient Nature inspired (hyper-organic?) forms can finally be applied to engineering applications.

Two of the recent examples of this kind of bottom up design that the Hyperganic team has presented publicly are their heat exchanger element, printed by Heraeus in copper and their rather impressive-looking cutaway rocket nozzle showing the internal conformal cooling channels.

Diagram showing how the Hyperganic algorithm takes several inputs to generate efficient, Natural-looking forms, including a copper heat exchanger and rocket nozzle (cutaway to show internal structure) (image sources: Hyperganic via 3DNatives and All3DP)

Hyperganic hasn’t publicly announced future support for multi-material printing, but there is evidence that they are taking it seriously. In an interview given at the Singapore National Additive Manufacturing Innovation Cluster’s Design for Additive Manufacturing conference, Hyperganic CEO Lin Kayser remarked,

“The one critical area that needs to evolve is multi-material printing involving metals. Once we can print metals reliably in a mixed workflow with insulating materials, the entire realm of electromechanical devices will open up. Printing of electric motors and actuators will be a complete gamechanger, because this functionality can be included in objects, essentially for free.”

Multi-material design seems to be an extension of Hyperganic’s existing capabilities that could require only minimal effort to put in place. They claim to be able already to control print parameters at the voxel level, and the voxel resolution can be defined according to the printer’s resolution. From a software and processing perspective this may be all that is necessary to distinguish voxels of one type of material from another. It’s a different story to predict real-life material behavior, especially when dealing with material interfaces, but this is an issue obviously not unique to Hyperganic.

Hyperganic’s software is not available for purchase. Rather, the company is making revenue-sharing partnerships with companies that need Hyperganic’s voxel engine for specific applications. This might be an attractive choice for small companies that have ideas with big market potential but not a lot of cash lying around to pay huge site license fees for software.


ParaMatters is a California-based company developing a software, called CogniCAD, that offers both generative part design for new components and topology optimization for existing CAD models. In addition to the unique capability to automatically optimize porous internal mesostructures according to applied stress, they also publicly advertise a multi-material topology optimization feature. In a two-part LinkedIn article, company co-founder and CTO Dr. Michael Bogomolny walks the reader through an example of multi-material part optimization, showing that a hypothetical part containing optimal allocations of aluminum and titanium could be made to be 2 kg whereas an optimized part made of aluminum alone was about 5 kg. In this case, not only was the hybrid part 60% lighter, it was also slightly cheaper ($235 vs. $250 for aluminum alone).

The original Y-shaped bracket (top left) was topology optimized for minimum compliance in aluminum alone (top right, grey contour) and a theoretical aluminum/titanium composite (bottom, red = titanium, grey = aluminum) (image credit: ParaMatters via LinkedIn)

CogniCAD is at its core a suite of proprietary functions that analyze and optimize part designs based on finite element analysis of stress distributions for given load and compliance constraints. For such analysis and optimization, models must be discretized into meshes with small volume elements, and this discretization process is similar to voxel rasterization. Based on Dr. Bogomolny’s articles and the ParaMatters website, it’s evident that CogniCAD already has the capability of assigning various materials to mesh elements and already makes some assumptions about the interfacial strength of material combinations. It’s not clear if the software can readily export the 3D discretized meshes as voxel-based models containing material information for printing, but it would likely not be great challenge to add this feature.

At the moment ParaMatters offers CogniCAD 2.1 as a cloud-based platform where users upload pre-designed CAD models to optimize according to various selectable parameters. In addition to the aforementioned mesostructural and multi-material optimization functions, vibrational considerations can also be used to optimize a model in CogniCAD. The company even claims to be able to optimize models without explicit load conditions being given, based on materials chosen, boundary conditions, and presumably fastener size (implied by fixation hole diameter constraints). The optimized models can be downloaded in STL or STEP format. Users can either pay per design or pay for an ongoing subscription to the part optimization service.

Additive Flow

London-based Additive Flow has recently emerged from stealth mode. They differ from the others on this list not only by offering an explicitly multi-material oriented software plugin (for Rhino CAD) and a standalone Windows application and also by offering consulting services for multi-material component design.

Thanks to our common goal to enable industry-relevant multi-material AM, Aerosint has been in contact with Additive Flow for some time now. To provide a bit more information, we asked Charles Fried, CTO and Co-founder of Additive Flow, for an explanation of their technology’s unique features. Charles writes:

“Additive Flow was founded to take the benefits of Additive Manufacturing to the next level. We create intuitive physics-driven Generative design software for multi-material and multi-property design. Additive Flow’s optimisation tools are relevant across the development and manufacturing value chain.

Additive Flow can handle the manufacturing capabilities of single or multiple materials, with custom parameters, and run real-time automated simulations and optimisations to provide rapid recommendations and trade-offs in a generative design process.

This holistic workflow enables our customers a broader exploration of design options in a fraction of the time, reducing the need for iterative steps: accelerating product development for better more sustainable products.

We often get asked what is different about Additive Flow in comparison to other generative design platforms. Our software itself offers many features not available elsewhere such as setting parameters for single materials, compatibility of multi-materials and their ability to be manufactured. We provide a clear comparison of single to multi-material so both constrained and unconstrained design engineers can have more choice in materials, geometry, applications to meet their technical requirements. Those who have an immediate need to apply functional gradients to achieve a leading-edge design and performance over competitors are talking to us as we can generate the machine code for this unique process.

We are working with a number of innovative, established hardware OEMs on technologies ranging from cold spray, laser metal deposition (LMD)/cladding, LPBF and electron beam melting to fused deposition modeling (FDM) and Binder Jetting, allowing us to tune what is possible for each technology and material type. We’ve already been working extensively with cold spray and LMD, for example, to develop approaches to join dissimilar materials with appropriate transitions while avoiding intermetallics.
Additive Flow unlocks R&D and Manufacturing in a seamless process. Companies are constantly looking for a competitive edge or working out how to improve a product’s life cycle, drive down development and production costs. To achieve these goals and produce a part or product that exhibits a performance improvement over previous iterations is where Additive Flow steps up the game in Generative Design software.”

Screenshots of Additive Flow’s FormFlow software show stress analysis (top) and corresponding discrete multi-material allocation (bottom) for a geometry-optimized Alcoa bracket

Additive Flow provides its software via subscription licenses, and provides support packages to help its customers achieve their innovation and production goals. The team also offers application engineering services to their top-tier customers for special projects.

The not-so-far, multi-material future

At the moment, the commercialized multi-material AM processes are primarily polymer-based, slow and best suited for prototyping — not for production. Multi-material AM of functional parts from polymers, metals, or ceramics is largely unproven, primarily because technologies that enable it (like Aerosint’s) are few and still in R&D or pre-commercialization stages.

Fortunately, we don’t need multi-material AM to be fully mature in order to begin predicting multi-material part behavior. As alluded to in the sections above, knowledge of material-material interfacial strength should be included as input parameters in topology optimization algorithms so as not to unrealistically overestimate the mechanical strength of a composite part.

Several multi-material forming methods (both additive and non-additive) exist that can help us determine reasonable estimates for these parameters. Two-component injection molding can be used to produce dual-polymer parts. Metal-metal interfaces can already be created in parts using cold spray or laser cladding technology (technically AM technologies). Finally, ceramic-ceramic interfacial behavior can be measured in components produced by established slurry casting and sintering processes. Mechanical characterization data from parts produced using these methods should provide a good starting point for incorporating interface behavior into topology optimization algorithms.

When multi-material AM techniques mature, process- and material-specific interface behavior can be used to update the part design and optimization software so that mechanical property predictions are even more accurate and relevant. Through this iterative manner both AM software and hardware will improve in terms of predictive capability and physical reliability.

There will be some fundamental limits to the range of material combinations possible, so we must remain realistic and careful in our choices of materials. Ultimately physics — not humans — will determine which combinations can be made to give robust interfaces or gradients. Still, some material combinations have already been proven surprisingly feasible, such as copper alloy and tool steel in a multi-metal LPBF process developed at the Fraunhofer IGCV, or copper alloy and an undisclosed, high-strength superalloy for building rocket nozzles.

Multi-material AM may find its broadest industrial use with simple (but extraordinarily useful!) material combinations. One good example that could have broad application is the combination of high and low carbon steels. Case hardening is a traditional post-processing step in which steel parts are treated to increase the carbon content (and thus the hardness) of the surface of parts while retaining a tough, low carbon interior. Multi-metal AM techniques could reproduce this mechanical behavior in situ with little doubt the two steels could be co-processed. An analogous approach applied to polymer- and metal-matrix composites (as the high-carbon steel analogue) combined with their respective pure matrix materials (as the low carbon steel analogue), could yield function- and cost- optimized parts that are very likely to be successfully processed in AM.

Final thoughts

Multi-material AM is still young but growing up fast. When it matures, the software innovators profiled here will likely be ready to provide the manufacturing industry with the needed design approach and solutions, and traditional CAD software companies will be scrambling to catch up.

Like them, we at Aerosint are confident that many useful material combinations will prove processable in AM and will become invaluable for a whole host of applications. Multi-material AM will keep developing because geometric freedom is only half of the promise of AM. Truly optimal engineered devices will be those in which both form and material composition follow function.

by Kevin Eckes, Ph.D., R&D/Applications Engineer

More To Explore