Process

To start this assignment, first I had to choose a shape I wanted to 3D print. I chose three to test: a flexi cat, a bird whistle which sat in an ornamental spherical shape, and a lego man figure. I measured the size of the capsule with calipers, and scaled the files (taken from Thingiverse) to a safely small size to test print in the Bambu slicer, before slicing the file and running a test print with a Bambu printer. When they had finished printing, I noticed that only the lego man and bird whistle were functional, but due to the small sizes of the moveable prints (I had underestimated how much space was in the capsules), neither were moveable and also the flexi cat did not print properly at all.

Test prints, small scale

I scaled the prints up to reassess whether the prints could be viable in a larger format, and to better fill the capsule volume. Upon reprinting, I noticed the bird whistle no longer printed properly, so would need supports if I chose to print it at this larger scale. I also noticed that the flexi cat could move some in the torso, but the tail was not properly moving. The lego figure worked perfectly, so I chose this figure for my project, and also knew that the printing settings on the Bambu printer for this print were functional.

Test prints, larger scale

Therefore I printed two more lego figures with the Bambu printer, although this effort took more than one attempt as twice the prints failed due to the arm rotating during the print. Notably, the Thingiverse uploader had written that this print does not require supports despite there being a slight (but gradual) overhang to print the forearms. I trusted their process, so opted to not use supports for the Bambu printer, although supports at the arm likely could have prevented the two failed prints due to arm rotation.

Chosen lego figure print fits perfectly into capsule

For the second 3D printing process, I used my past experience to save some time (with permission!). I have worked fairly extensively with SLA printing, using Anycubic printers throughout my master’s thesis and Formlab printers during a six month internship for my master’s program. Based on my experience with SLA printing, I knew that the resin prints would be far too rigid to achieve motion, and the print would either fail or need complex internal supports hindering motion. Therefore, I opted to use two different FDM printer types to complete this assignment based on what was available in OEDK: the Bambu printer to print three figures, and a Prusa printer to print the remaining two figures. Since I had working parameters for FDM printing using the Bambu printer, I started by simply replicating these parameters into the Prusa slicer.

Adjusting Prusa slicer settings to match Bambu printer

The first print I attempted quickly failed as the filament was misaligned in the nozzle of the printer, so I tried on a different printer, trying to print two figures next to each other. At first, it appeared this print would fail as well, but with advice from a classmate, I slowed down the print speed, and also increased the temperature of the nozzle and the speed of the fan. The first layers looked questionable but not failed, so I gave the print more time and ultimately one of these two figures appeared promising so I allowed the print to finish, with one figure working (other than very slight imperfection on just the first layer or two, and the other figure having one arm printing irregularly in the first few layers, leading to some ‘spaghetti-ing’ of the structure. I only kept and submitted the better of these two.

Slight imperfections in first couple layers of functional Prusa print (seen in upper back, arm layering seen in all prints)

Spaghetti-ing of arm seen in failed print

To print the second figure, I had to troubleshoot a bit as the prints kept failing on different printers. Even adding supports failed to make the prints work, indicating a problem with bed adhesion more than with the print files themselves. I consulted with Hayden, who gave advice to increase bed adhesion by adding a layer of glue to the bed pre-printing, and slowing the first few layers to 70% speed before returning to 100% speed after. He also suggested I add a brim to the file pre-slicing, and also using supports only to the bed. With all of this advice, this print worked, yielding my second functional figure from the Prusa printer. Due to the use of supports, the supports had to be peeled off yielding the completed figure.

Using glue stick to improve bed adhesion

Print on support structure

I forgot to take a picture of the final printer post-printing to demonstrate that the workspace was clean, but by removing the prints (and test lines) from the printers, I left the printers ready for the next user.

Removing print bed to clear of print and test lines in an earlier print

Cost Model:

Cost Type	Cost	Price	Source	Quantity	Total
Materials	PLA filament (Bambu Lab)	$24.99 /kg	Bambu Lab	.1 kg (overestimate)	$2.50
	PLA filament (Prusa 3D)	$29.99/kg	Prusa 3D	.1 kg (overestimate)	$3.00
	Toy Vending Capsules	$2.45/ 50 ct	Amazon.com	5 capsules	$0.25
Labor	3D printer operator	$21/hour	ZipRecruiter	1.5 hour	$31.5
	Prototyping Engineer	$36.5/hr	Indeed.com	1 hour	$36.5
Overhead	Facility Cost (Machine Time)	$40/month	The Maker Barn	1 month	$40

Total: $113.75/5 prints = $22.75

The cost model for 3D printing involves many tradeoffs as one of its most significant features. The cost of filament is rather negligible, so the most expensive aspect is labor when not purchasing printers, although many prints can be run in parallel, so one operator could run a considerable number of prints at the same time, leading to a relative low labor costs. However, in order to run this many prints in parallel, there is a need for multiple 3D printers to be used at the same time. In a facility like The Maker Barn, there are likely policies preventing use of too many printers by the same individual. For 3D printing, the best financial choices are completely dependent on the scale of production one aims for. For industrial applications, it is likely wisest to pay a high upfront cost and buy many 3D printers that can be run in parallel by relatively few operators. As a result, the labor costs would be expected to decrease considerably despite a large scale of production, which over time will result in net savings in comparison to running fewer printers and thereby attaining smaller production scale, which also involves paying for ‘unused’ time from the operators while they wait for prints to run. The number of printers per operator could be optimized in such a setting, and by balancing all of these factors, the cost per item produced could attain a theoretical minimum. With this process optimized, the costs of other components of the current cost model would decrease per item produced to a negligible quantity due to the lack of a need for further prototyping once the production parameters are optimized.