Calculating the cost of additive manufacturing

Jason T. Ray is the managing director of J. T. Ray & Associates, an advanced manufacturing industrial strategy consulting firm

This body of work is a direct response to requests, received by the author—additive manufacturing industrial strategy consultant Jason T. Ray—for a full breakdown of the direct and indirect costs associated with additive manufacturing (AM), most notably in response to his article entitled Economic Benefits of Additive Manufacturing in Aerospace. To address this issue, Jason runs the numbers so that readers of Disruptive Insights can see how properly allocated costs add up. The aim is to provide a comprehensive overview of activity-based costing for the end-to-end AM process.

Always start with why?
Why is it so important to understand how costs in the (AM) process are allocated? If you are a buyer, this understanding will help you dissect quotes from service bureaus and ask educated questions during the negotiation process (i.e. what do you anticipate the machine run time to be and will my part be the only one produced in the build?). If you are an AM facility manager, it is paramount to understand the cost drivers associated with any manufacturing process.

Calculating this breakdown and understanding cost levers is the first step in finding innovative ways to enhance your operation's efficiency. Furthermore, it will also help users to assess the true value in the incremental machine improvements (such as print speed, multiple lasers, sensors, surface quality, reduced post processing, automated powder handling, etc.) that are regularly being introduced by industrial AM hardware vendors. How will that 20 percent increase in print speed or integrated powder handling system really impact your bottom line? What are the most important innovations coming to the AM market that will drive efficiency in your operation?

Breaking down the costs
For this exercise, I will look at a fictional satellite bracket manufactured using the direct metal laser sintering (DMLS) process out of titanium (TI64). The purpose of this is to provide a focus, but the issues raised are relevant to any specific AM application.

The costs associated with most additively manufactured components can be broken into four distinct sections—design, production, post-processing and qualification. Whether each step in the process is performed at one location, or across a value chain of internal and external suppliers, the numbers will add up to the total all-in cost for the component. It is important to recognize that there are also overhead costs (utilities, rent, G&A, shipping, etc.) associated with each step of this process, as well as consumable items (inert gas, filters, protective equipment, etc.) and we will address these costs at the end.

Note: The process below is often iterative, requiring multiple run-throughs and revisions to achieve a successful part. If this is the case, the costs associated with the number of iterations (design/redesign, production, post-processing, qualification—repeat) it takes to achieve a qualified, functional component should be amortized across the total number of final parts produced. This is akin to allocating R&D/product development costs properly.

Design for AM
Design for AM is the first step in the process. Components should always be designed—or redesigned—for AM specifically to maximize the greatest end-use benefits (reduced weight, increased strength, component integration, etc.). Due to the novelty of AM and high upfront investment costs required to buy the machines, the use cases often turn into a hammer looking for nails, rather than a cost-benefit driven process. When this happens, emphasis on redesign is reduced and companies end up producing components that were originally designed for traditional manufacturing methods and thus only capture a fraction of the end-use benefits associated with AM (i.e. supply chain efficiencies). So, if you are wondering why your company's recent AM system purchase isn't paying off the way it was expected, it is time to look at how you are employing the technology.

Note: Software is going to play a huge role in reducing the costs associated with the design process over the next two years. Stay tuned!

Bracket design assumptions (SWAGs):

Design Time:

1 month

Fully Burdened Design Engineer:

US$120,000 per year

SolidWorks Premium:            

$7,995 (one-time license fee)

SolidWorks Annual Subscription:

$1,995

Software Useful Life:

10 years (assuming new products or solutions requiring additional investment) $7,995/10 = $799.50 per year

Note: SolidWorks was chosen for this example because of the availability of cost numbers and ease and is therefore indicative. However, the author is aware that SolidWorks (like many currently available CAD software suites) likely does not have the capabilities required for this sort of topology optimization for AM. The bracket in the picture above was designed by Altair ProductDesign using Altair OptiStruct for topology optimization and structural analysis and solidThinking Evolve for CAD generation and rendering.

Bracket design cost calculations:

1 Month of Labor Cost:

$120,000/12 months = $10,000

1 Month of Software Costs:

($799.50/12 months) + ($1,995/12 months) = $232.88

Total Bracket Design Costs:

$10,000 labor + $232.88 software = $10,232.88

If this cost is to produce only one bracket, then the allocated cost is as shown above. If it is to produce multiple brackets, assuming that only one bracket can be produced per build due to size restrictions, it should be allocated across total number to be produced (assume 20 for the example below).

Per Bracket Design Costs:

$10,232.88/20 = $511.64 per bracket

Note: When looking for existing components to redesign for AM, it is important to consider the quantity of parts as this will play heavily into your business case or cost benefit analysis.

Production
Production can be broken down into the following areas: materials, setup (design and machine), machine run time, and final build plate removal. An area where some service bureaus get in trouble is not properly accounting for fail rate. An additive build can fail for one of a hundred or more reasons, rendering the final product useless. This risk must be accounted for across each of the production categories to ensure the shop can absorb the costs associated with inevitable failures. The fail rate calculation will be addressed at the end of this section, however this may also impact every remaining process if failures are not caught early.

For instance, if the manufacturer only finds out that the part failed in the final QA/qualification step, each of the steps leading up to it are lost dollars as a result of the scrapped part. This is why in-situ process monitoring is so important for the future of AM—to catch failures as quickly as possible and stop your machine/post-process to avoid unnecessary costs. This is similar to traditional operational efficiency tactics where QA is moved forward in the manufacturing value chain to help cure bottlenecks and reduce costs.

Materials: The cost of powdered metal is significantly higher, in some cases an order of magnitude higher than bar stock, for several reasons including: energy intensive production process (gas atomization); purity; and required powder sphericity and particle size uniformity. The cost for powdered TI64 can range from 150 to $400 per kilogram. Two things to consider in determining material cost are the component's weight and the scrap rate of powder that cannot be reclaimed for follow-on builds.

Material assumptions:

Price for TI64 (per kg):

$200 (ballpark assumption)

Bracket Weight:

1 kilogram

Scrap Rate:

10% (accounting for both support structures and sintered powder particles siphoned out at the end)

Calculation:

Total Bracket Material Cost:

(1kg x 1.1) x $200 = $220

Note: There may also be some material qualification costs outsourced to a lab, however for the purposes of our study, we will assume that the material provided came with these tests already included. As a buyer, relevant questions to ask your metal powder provider should cover third-party verification of particle size, chemistry, flowability, etc.

Setup (design and machine): Even though our bracket may have been designed specifically for AM to capture the maximum end-use benefits associated with the process, this does not mean it was designed for additive manufacturability. Thus, part of the setup cost is associated with build area/platform management. Our bracket may require specific support structures or orientation to ensure the build is successful. Each AM machine has different capabilities, so this process usually has to be done by an experienced operator. There is also a physical labor cost associated with setting up the machine with the right powder and new build plate to prepare it for the build.

Setup assumptions:

Setup Design Time:   

1 hour

Setup Machine Time:

1 hour

Fully Burdened Operator:

$120,000 (2,000 hours per year)

Calculation:

Total Bracket Setup Cost:

2 hours x (120,000/2,000 hours) = $120

Note: This assumes our bracket is the only part being produced in the build. If there were multiple parts being produced, the physical machine setup time should be spread across each part.

Machine Run Time: The hourly rate on your AM system is determined by the upfront and maintenance costs associated with it (to include a cost of capital-hurdle rate or interest rate), your expected capacity utilization rate, desired profit and your payback period or useful life. This is one area where traditional machine shops adopting AM technology end up in trouble. 3D printers are not CNC machines. The useful life should be dictated by the product iteration cycles and time to obsolescence or relative inefficiency.

New AM features and lower cost machines are coming out roughly every six months. It is safe to assume that every 24 months you will need to buy new technology to remain competitive. Companies that underprice their machine run time will end up in a tail chase situation where their competitors are upgrading machines and able to offer lower prices based on new capabilities/efficiencies (i.e. build speed). This will make it even harder for the underpriced operation to catch up and generate an ROI on their initial investment. There is a careful balance here between maximum capacity utilization while charging a high enough price to justify your investment.

Machine run time assumptions:

Upfront Purchase Cost:

$750,000

Yearly Maintenance Contract Cost:

$10,000

Payback Period:

24 months

Hurdle Rate:   

5% (this is the interest on the loan, the hurdle rate would be higher if we were looking for a profit as most service bureaus are).

Asset Utilization:

80% (this number should be changed to reflect your actual capacity utilization. 80% is near full utilization for a DMLS printer when accounting for maintenance and turnaround time).

Bracket Print Time:

40 hours (We will assume that the bracket is the only part on the build plate. Print times vary from a few hours up to 150 hours, so keep in mind how long this process can be and then think about cost of a print failure in the 149th hour).

Calculations:

Monthly Machine Cost:

((750,000 x (1 + (5% hurdle x 2 years))) + 20,000)/24 = $35,208.33 (This is a piece of a somewhat simplified NPV calculation that errs on the high side.)

Available Monthly Machine Time:

(30 days x 24 hours) x 80% = 576 hours

Hourly Run Time Cost:

$35,208.33/576 hours = $61.12

Total Per Bracket Run Time Cost:

40 hours x $61.12 = $2,444.80

Final Build Plate Removal:    

When the build finishes, the part must cool and then be removed from the machine. This requires an operator cleaning off and recapturing unsintered powder and in some instances removing the part from the build plate.

Build plate removal assumptions:

Removal and Powder Recovery Time:

1 hour

Fully Burdened Operator:

$120,000 (2,000 hours per year)

Calculations:

Total Per Bracket Removal Costs:

1 hour x (120,000/2,000hrs) = $60

Fail Rate—to understand the impact of the failure rate, let's first tally the costs.

Production total:
$220 Material + $120 Setup + $2,444.80 Machine Run Time + $60 Build Plate Removal = $2,844.80 Per Bracket

Notice how much of this number is driven by machine run time. Thus, print speed—often driven by the speed of the recoater during the build process—is a significant metric in metal AM. Moreover, how you orient your build matters!

Assuming you discover failures after build plate removal at a rate of 20%, this is the premium you need to add to your costs to account for the scrapped parts.

Fail rate premium:
$2,954.80 x 1.2 = $3,545.76

This fail rate has a significant multiplier effect. Some service bureaus account for this in their profit margin, but when running a printer internally, managers must understand and account for the potential fail rates in both cost and time projections.

Post-processing
Post-processing requirements depend on the geometry of the component and how well it is designed for manufacturability using AM. This is a key indicator of how well your designer understands the intricacies of the AM process and specific machine capabilities.

Parts requiring a significant amount of post-process machining will be more expensive because of the CNC programing required to set up the tool paths to process the part. This cost has much less of an impact when producing larger lot sizes. There is also a level of risk associated with this, as the wrong program or tool path can render an already expensive AM component damaged and unusable. This is an interesting area where hybrid additive and subtractive machines will come into play in the future, I believe.

Post-processing includes heat treating, machining and surface finish processes. To spare you a long explanation and redundancy, for our bracket example, we will assume that no post-processing is required based on the component's design and end use. However, if it were necessary, each of these processes follow a very similar calculation to that of the production steps outlined above: setup, machine run time and removal.

Qualification
The process required to qualify a AM component varies drastically based on its end use and thus the cost can fluctuate from 0 to over $10,000 USD per part. For complex commercial aerospace components like the LEAP fuel injection nozzle, which will all require FAA approval, the non-destructive testing process is extensive. It includes full traceability of each component, source materials, and a computer tomography (CT) scan with a hands-on engineer review of that scan in its entirety to ensure internal geometries are accurate and required densities (i.e. no porosity) are achieved throughout the entirety of the part.

We can assume that our bracket example will not require this sort of inspection process, however, this process can add significantly to the overall cost of each component with both labor and CT scan allocated run time costs. Now think about the impact of recognizing a component failure in the 11th hour, that's a lot of money and time spent just to scrap the final part.

On the lower end of the qualification spectrum, a component may only need to meet certain strength and geometric requirements, so breaking some sacrificial tensile bars strategically placed throughout the build and taking a few measurements might suffice (although this will increase your scrap rate on the material cost). For the sake of wrapping this up, we will assume for this bracket that some simple dimensional checks will suffice and thus the qualification costs are negligible.

Note: Soon in-situ process monitoring from strategically placed sensors that track temperatures, melt pools, oxygen levels and laser strength will provide real-time analysis of the build process. This data will only be accepted as a form of quality verification when a sufficient baseline of data from successful and failed builds exists, proving its validity in a statistically significant fashion. However, these sensors should be able to identify failures much earlier on in the process.

Overheads and consumables
There will be numerous items that fall into this category depending on the AM process being used, facilities, support staff and liability. Overhead costs (utilities, rent, G&A, etc.) associated with each step of this process are fully dependent on the individual operation and it is important to assign these costs appropriately. If these are viewed as sunk costs versus allocated capital with an opportunity cost, there is a good chance for inefficiency.

Consumable items (inert gas, filters, protective equipment, etc.) will again vary depending on the utilized AM process. For our bracket example, being a metal component, there would be more energy used in the DMLS process and inert gas required for the build chamber. These items all add to the cost of the final product and should be built into your cost algorithm.

Conclusion
We have covered a lot and if you are still reading, thank you! There is much to consider, but I believe the most important takeaway from this overview is a comprehensive understanding of how to properly allocate the activity-based costs associated with the end-to-end AM process. This will help you make a cogent front-end business case for the technology's use and enable you to identify the largest cost levers, so you can drive efficiency and targeted innovation in your organization.

For more information on the topic, check out the infographic provided here: Making the Business Case for Additive Manufacturing: A Manager’s Guide.

This article was originally published by Jason on LinkedIn December 2016, and is republished here with permission, ahead of an ongoing collaboration with Disruptive Magazine.

About Jason T. Ray

Jason is the managing director of J. T. Ray & Associates, an advanced manufacturing industrial strategy consulting firm, working with Fortune 500 companies and leading manufacturers.