8.4 AM Economics

Feedstock

For metal PBF, the feedstock is a powder. Traditionally, the metal powder is made via an atomization process whereby the metal is made liquid by heating and the molten droplets fall into a tall chamber. The droplet is broken up by a variety of methods which cause many smaller droplets to form before the they cool and solidify as they fall through the chamber. The output of the process tends to be a spherical powder with a Gaussian distribution of diameters. Not all metal powders are made this way, however, and there are many variations on the concept. Atomization is done via gas atomization, water atomization or plasma atomization which vary in the medium used to break the liquid metal droplet and the raw materials to create the liquid metal. Other forms of powder production exist and have been used in powder metallurgy such as Hydride De-Hydride (HDH) and mechanical methods. In polymeric PBF, the powder can be synthesized or mechanically ground which gives very different characteristics in terms of roundness and size distribution. For our purposes in this chapter, the size distribution will be a particular focus because it drives the yield and therefor productivity that process to make feedstock.

Powder Size Distribution (PSD) is a key factor in any powder production process. This is mostly because the printing process has prescribed a fairly narrow PSD to be used in the process, while atomization produces a fairly wide PSD, regardless of which method is employed. The selection of PSD contributes to surface finish, the ability gets a desirable Packing Factor in the bed, the ability to move the powder and ultimately cost. PSD can contribute to cost in the powder price, the printer performance and the downstream processing so it is a pervasive cost driver. It is also important to point out that size is a factor in the ability to move particles with smaller particles flowing more poorly than larger, which is why roundness or sphericity is often discussed. Spherical particles flow better than non-spherical particles, so when the requirements call for freely flowing, narrow PSD and small particles; having spherical particles helps. Naturally, having fewer smaller particles also helps as well as having particles that can be moved versus having “high flowability” are important specifications. If over specified, the powder costs more than the AM process truly needs to produce a desired output.

AM Process

Within the AM process, the speed of creating a layer and melting the material is critical to productivity. The layer creation step is influenced by the speed at which the powder can be moved into place, the utilization of the powder (i.e. how many times it can be re-used) and the thickness of the layer. In PBF, for example, the scanning step where the laser (or electron beam) is melting the material and thereby creating the part is influenced by the number of lasers and the part thickness. Another nuance is the addition of lasers to create a larger bed size where more parts or larger parts can be made. All of these factors affect the productivity of the printing process. They also then impact the upstream and downstream processing as well.

Print Speed

Print speed is often mentioned when discussing productivity or comparing various AM processes. It is important to take this into context as increasing the print speed alone may not increase the speed at which a part can be made and, in some cases, a faster print speed can be more expensive at the part than a slightly slower speed. It is logical to assume with expensive capital expenditure (CAPEX) items, like 3D printers, that faster would be better, but this often obfuscates several important factors. There is a difference between the print speed, the effective print speed, and then the actual cost to produce the component. We will discuss:

• Actual print speed: Speed at which the printer makes a shape
• Effective print speed: The resolved speed of the printing when considering the print time and required steps to make the print useful

Finally, the final part cost is the ultimate metric if the productivity of the entire process has been improved or not.

AM is a series of processes, and those processes must be matched or optimized to get the maximum benefit. Inherent in these examples has been the ability to make thicker layers. We will now look specifically at the economic considerations of thicker layers in a series of processes.

Printing thicker layers can be more difficult than it sounds. We’ve already mentioned that thicker layers can have implications on the powder input. Thicker layers will also impact the processing downstream because of the lower resolution. If the part is extremely detailed, thicker layers may not work but for larger parts or parts without a high degree of detailing, using thicker layers is a means to making more parts faster.

By increasing the layer thickness, we effectively reduce the print time. Going from 30-micron layers to 60-micron layers isn’t exactly a 50% improvement, but it is close. Thicker layers only impact the time during printing when the part is being scanned or made. Recoating time is unchanged as is the set up and cool down time.

The figure below shows the resulting calculation to examine the impact of thicker layers on part cost due to reduction in print time. Increasing the layer thickness by 50% leads to a reduction in print time and consequently part cost of approximately 20%.

Figure 8.8 The calculated impact of layer thickness on part cost.

The next focus then becomes the cost of the powder because we have shifted our part cost contribution. The figure below shows the new cost implication as escalating the impact of powder as the downstream processing is more or less unchanged.

Figure 8.9 Cost driver impact from a 30-micron layer baseline to a 60-micron layer, and same powder PSD and cost/kg.

Machines

As the printers or machines are the single biggest factor in the cost of making parts, we will consider them first. With some AM technologies, the printer comes with the peripheral equipment required to finish the part. Polymer PBF, for example, does require some post printing treatments, but due to the maturity of the technology, there are multiple suppliers and automations available. Metal AM, which is less mature by comparison, has a very immature supply chain and is somewhat scattered. We will examine two aspects of machines in a business case: batch vs. series production and operational cost.

In production, the processes must be aligned in a fashion to optimize the part flow. Through this optimization, the producer can get the most from the operational uptime of the equipment. AM can be thought of as a series of batch operations. Batch operations are where a service is performed on a group of parts at the same time, like heat treatment. Several parts are placed in a furnace, and it runs for hours. The cost of running the furnace is the same whether it is full of parts or has one part.

Clearly, we would be wise to optimize the maximum number of parts we need with the chamber size of the printer and the capacity of a furnace. When these are matched well, we can see a benefit close to series production called batch series. Batch series simply describes where two batch processes are matched so that they in effect act like series production. Series production is more closely related to the original Ford model of assembly line efficiency. The biggest difference here is that we have a printer which is good at making many different designs so keeping it running may mean making one part in the first week and a different part (same material) the next. This is most commonly referred to as “agile” manufacturing where capital equipment is capable of making many different things as they demand arises.

The economics or operational effectiveness of AM equipment and the legacy manufacturing equipment are essentially the same. The Accounting of the depreciation of the capital equipment sets in place the hourly cost of using the equipment. The cost of using the equipment then drives how we can design the parts to balance cost and performance.

Beyond the hourly cost of the machines, the next factor is being able to deploy them such that our processes are aligned as much as possible to seek efficiency. Aligning our production so that part flow from the printer to the furnace is an example of lean manufacturing. Lean manufacturing is a methodology that focuses on minimizing waste within manufacturing systems while simultaneously maximizing productivity. In some ways, lean and agile manufacturing have opposing requirements, but both promote minimizing waste and focusing on productivity. The difference is perhaps that lean would prefer to make the same widget every day whereas agile responds to the market demand. Agile is then more difficult to predict and therefor harder to manage in manufacturing.

While we have focused on the AM machines, we have also discussed the need to align production capacity with downstream processes. This is more prevalent in metal AM than polymer AM as metals are more exposed to supply chains. The premise of our discussion and examples however are the same. It will be necessary to match the production output of the printer to the stress relief furnace to the HIP vessel to other finishing processes. The ability to do this effectively will drive a decision on what can be done in house more efficiently than outsourcing where you seek market efficiency.

Process-by-Process Discussion: Emphasis on Metals

PBF and DED with powder have the majority of cost in printing and post-processing. Materials, while perhaps more expensive than typical mill products are used so efficiently, most of what you print with you use and remove very little. The AM equipment is expensive, so the depreciation drives the overall cost. Using the printer very productively is key. A key value proposition is making very detailed, very near net shape products either from the platform up or adding on to an existing product form or repair.

Sheet lamination has minimal printing costs partly because the print speed is higher but also because the feedstock (thin gauge sheet) is expensive, and the process provides details at the expense of material wastage. The printing and machining are coupled leaving post-processing to traditional metal working and final finishes.

Binder jetting has more post-processing intensive than other processes because it needs to sinter the part in order to achieve full density. The material cost is somewhat artificially inflated because we have normalized the whole cost to 100%. Materials in BJP are typically less than laser powder bed fusion, for example.

DED with wire is very efficient at printing. It will always require a heat treat for stress relief, perhaps multiple times, and machining due to its inability to make details. Wire can be an expensive product form due to the diameter that is typically required is around 3 mm. Wire is made from bar stock and drawn or swaged down which is a lengthy process.

Direct Extrusion is quite fast and not very expensive, so for putting a lot of material down, it is very efficient. The feedstock is more like bar stock or could be waste material, so readily available and not expensive. It does require similar post-processing to DED + wire due to its inability to make near-net shapes.

Cold spray is very fast and very affordable. It does not make near net shapes so additional machining will most likely be required but the deposition rate can be tailored to produce acceptable surfaces. In some cases, a heat treatment could be necessary as well. The powder cost for cold spray can vary wildly depending on the chemistry or alloy from very inexpensive to more expensive than laser powder bed fusion, specifically in titanium.

Materials

In some instances, the material will help decide the AM process needed. In other cases, where equivalent or superior performance is all that is required, a superior material could be used with better performance and no effect on cost. In some cases, like PBF, the material cost as a percentage is low. In this instance, evaluating alternative materials means getting higher performance at negligible cost impact. Consequently, in Sheet Lamination where the material input is high, further investigation would be required.

Figure 8.10 Each of these canisters contain 10 kilograms, about 22 pounds, of powdered steel, used in the construction process with the EOS M 400 3-D metal printer. Each container is valued at $7,000. The 3-D printer also used soft aluminum or titanium powder to construct hard to find parts or design and build tools specifically for a project at PPB, MDMC. (credit: Marine Corps photo by Keith Hayes on DVIDS, Public Domain)

The choice of material can inflate the importance in processes like BJP and SL because the PSD and thin gage requirements are more expensive in titanium than they would be for example in steel which is more easily rolled to thin gage. However, this choice also drives the point in PBF and DED powder which is that titanium whilst having a high relative $/kg than other materials is <20% of the overall cost.

In consideration of the business case, it would be necessary to consider how the AM feedstock price compares to the legacy form price. This assumes material substitution is not a possibility. If the cost of AM feedstock, being a powder, wire or sheet is multiples higher than the legacy method, it begins to describe how productive the AM process is going to have to be as a first order estimation.

Accurately predicting the waste is another consideration. If the powder is converted to part efficiently, as in cold spray, the wasted material is quite low. In powder bed processes like PBF and BJP, powder may be re-used multiple times but for some reactive materials and polymers, the chemistry may dictate a limited number of uses before it must be scrapped. Wire or filament processes are typically very efficient at converting the feedstock into a usable part.

For AM processes that allow nesting of parts, there are details associated with support structures or sacrificial material that is used to improve the survivability of the part. In some cases, like the support structures used in PBF will consume material and then also require work to remove, there is a double effect. In other cases, where a plate is used in the part design, careful selection and placement of the deposition will help minimize the material wastage. Needless to say, when multiple material input forms are used, such as in DED, sacrificing the lower value material is typically preferable.

Digital

The digital component for AM encompasses a large catch-all of activity, as described in the chapter on the digital thread. First, it includes the designing for AM activities as well as the complete workflow associated. Generative design tools, topological optimization (T.O.), simulation tools are all separate components although some integration exists. From an economics point of view, all of these tools must overcome the cost of the tool in order to be productive.

It is easy to find the value in designing for AM, both MfAM and DfAM components. Without this view, it is unlikely for any value to be derived using AM where cost is the ultimate goal. Tools like generative design assist in finding the optimal shape for the given structural or thermal load. We find economic value not only in the higher performance but in the preservation of material as well. Topological optimization can then be used to further optimize the shape but can be used in other ways as well, where traditional CAD tools would be very laborious. Creating repeating or scalable non-solid sections like lattices would be very tedious and time consuming in CAD, whereas it is very efficient with T.O.

Of course, minimizing the design effort for new or re-design for existing applications, have significant economic benefits to the business case. These efforts are required and represent costs that will have to be recovered by the AM production to be of economic benefit.

Analysis and simulation tools are very helpful in achieving early print/build success. The time spent on the simulation tool and operation, however, is significant. So these tools are typically applied where series production is likely, otherwise, employing expensive tools might be more expensive than trial and error for a prototype.

Even with the optimization tools that we have, just as in conventional manufacturing, sometimes it is necessary to move forward with an imperfect solution as the analysis can continue to optimize but with diminishing value. Realizing when you’ve reached a point of diminishing value can often be difficult for technically driven organizations.

It is worth emphasizing again that designing for AM doesn’t stop at the printer. Providing location and data features to the design as it moves through post-processing is a key AM component, and further evidence that AM is a series of processes. Consequently, there is significant economic benefit when the design is optimized for the entire value chain, and not just the printer. The printer may not always be the most expensive piece of capital

Figure 8.11 Considering that AM is a series of processes, manufacturers and decision makers must realize that the printer may not be the most expensive part of the process. Here, a line of complex machines is arranged to produce advanced carbon materials. The investment goes far beyond the AM machinery and software. (credit: Modification of “Carbon fiber technology facility” by Oak Ridge National Laboratory/Flickr, CC BY 2.0)

Once the manufacturing processes are complete, capturing and interpreting the data from the build and inspection processes is also of importance. Through a series of dimensional and volumetric inspections, the manufacturer can determine if the part is capable of meeting the final requirements. This is no different than conventional manufacturing. Knowing when to apply non-value-added tasks like inspection is key, since they consume resources and time. Strong inspection practices throughout the process can ultimately reduce the cost of manufacturing a product, such as by immediately scrapping a part that is not going to meet the requirements before it is fully finished.

One last component of the digital perspective is the quality assurance (QA) and intellectual property (IP) protection. With digital files, QA and IP. protection are two sides of the same coin. With current tools, it is possible to pre-check the setup of a build to ensure what is to be made can be made. Additionally, tools that observe the build/print process can leverage machine learning to determine if human intervention is required or if the build should be terminated.

The build files themselves can be encrypted. There are several reasons for wanting to encrypt the build file:

• Protect the IP of the design which has economic value,
• Ensure that the wrong file isn’t used, and
• Ensure that the file to be built is what gets built and no damage or interference will occur.

People

Investments in people are usually of good value. In addition to building engagement with the team, investing in skills will yield a productive benefit. If anything, AM shows us that training and upskilling can be the difference between economic success or failure. Simply by looking at the role of design in AM, it has been shown numerous times how the design impacts the viability of AM.

It should be noted that investments in people should be made across the organization, not only in engineering or manufacturing. AM is disruptive so when an organization uses the same alphabet, they can more accurately and quickly communicate. The economic benefit to upskilling the organization is getting the design community to think and uses DfAM skills, but also engage and drive support from program management, accounting and supply chain as all of these organization can drive value in AM if they truly understand what is possible and not viable.

Outsourcing

If there is limited in-house expertise, outsourcing is a compelling option. It could be that a company specializes in design versus part manufacture. Even if the company is competent in part manufacture, should they be involved in every aspect of AM production? Integrating all of the processes into a single facility can be very expensive, especially when the minimum capacity for one process far exceeds the maximum capacity in another, as in the previous sintering furnace example. Additionally, adding in-house capability that will go underutilized increases overall costs without adding any offsetting revenue.

Logistically, being able to align the processing can add many queues. This could be critical when assessing the overall manufacturing time and working inventory. A company often pays for the material when it arrives, but isn’t able to receive payment for its effort until the part is built and delivered. The time in between requires the company to keep inventory, and reducing this time is strongly encouraged.

Ultimately, the decision to outsource or insource (keep everything in-house) involves several factors, but typically requires a balance between efficiency and quality. Access to customers, knowledge of materials, protection of intellectual property, or competitive experience could all be compelling reasons to insource a process. A company must know why it is keeping a process in house, so that if the rationale is no longer true at some later point, the company can consider outsourcing and make the correct choice.