Oh yes, we did.

“Oh yes, we did.”

For the past year we have been busy building, testing, documenting and refining the process of taking 3D printed parts and using “Lost PLA” burnout to cast for parts for more robust applications. The documentation is bordering 100+pages, with 20+ pages of brute force data. We will try to keep it simple, show off with a few shiny throwbacks, hopefully inspire ideas for the potential, and give some technical specs to boost the capabilities of those open source open hardware folks who love a good clean walkthrough. [click here for more information]

This is a rough sketch of how to assemble a mini blast furnace

This is a rough sketch of how to assemble a mini blast furnace

This design prevents the vacuum from sucking up molten metal if the plaster in the flask fails to seal.

This design prevents the vacuum from sucking up molten metal if the plaster in the flask fails to seal.

The sketches go through the simple breakdown of a furnace in basic parts and vacuum trap parts. Blast furnace information can be found here.  Any casting plaster can be used for when investing flasks for casting.

 

Load Kiln

Load Kiln

 

This is a generic burnout template, that works for most applications.

This is a generic burnout template, that works for most applications.

 

Clean burnout occurs ~1000-1300F

Clean burnout occurs ~1000-1300F.

 

Flasks are primed over a vacuum chamber before receiving the charge of molten metal

Flasks are primed over a vacuum chamber before receiving the charge of molten metal.

 

The test metal was scrap 6061 aluminum, and/or silicon bronze to ensure anyone could replicate the process easily.

The test metal was scrap 6061 aluminum, and/or silicon bronze to ensure anyone could replicate the process easily.

This is how parts look after they have been quenched, with no cleanup, simply rinsed with water.

This is how parts look after they have been quenched, with no cleanup, simply rinsed with water.

These parts yielded data about hole size requirements and edge cases. The goal was to quantify what was likely to succeed.

These parts yielded data about hole size requirements and edge cases. The goal was to quantify what was likely to succeed.

 

Casting Data

Parts can have clean interior corners, where CNC machines would fail to accomplish because of the cutter size. Self intersecting geometry is also not a problem. Edge case castings have been hearty with 13 fins space 1.6mm apart extending 15mm up and continuous for 40mm. This means complex geometry for cooling fins has little cost to prototype.The hard part is conceptualizing how volumetric shrinkage occurs. Basically the part will shrink ~2-3% depending on the alloy, but holes will get bigger as metal contracts from the side walls of the plaster. This means that parts need to be scale up ~2% while holes need to shrink by 2%. This allows parts to be well toleranced if machined afterwards.

The best part for testing the capabilities of any machine or process, thank you Loic.

The best part for testing the capabilities of any machine or process, thank you Loic.

A simple linkage cast separately and then assembled.

A simple linkage cast separately and then assembled.

Extremely complex parts that cannot be machined can easily be cast in production volumes allowing standard 3D print/cast parts to; withstand high temperature applications, parts have higher strength to weight ratio, parts can be custom bearing/bushing systems(when bronze is used), and parts can be used to create custom heat sinks (when aluminum is used).

The goal was to see if multiples could be cast simultaneously to minimize cost

Scalability for print to cast can minimize labor cost because sprue trees snap together like legos.

Scalability for print to cast can minimize labor cost because sprue trees snap together like legos.

Rapid manufacturing being applied to test injection molded screw caps [blue material is LDPE]

Rapid manufacturing being applied to test injection molded screw caps [blue material is LDPE]

Rapid manufacture of injection molds allows for even the smallest of shops to become competitive with standard injection molding. 3D printing adds ease and flexibility for companies to change their designs/molds faster and keep up with the demand.

More in depth information can be found here…  Rapid Manufacturing MK3

Cast bust of a 3D scan

Cast bust of a 3D scan

Tags: , , , , , , , , , , , , , , , , , , , , , , ,

13 Comments on Rapid manufacturing: Iteration and Industry

  1. Dave says:

    Fantastic write up! Thank you for documenting your process.
    @DJL_F3D

  2. tim says:

    I would suggest for a clean burnout of the can you need to move the kiln from 0deg f. to 600.F holding for 2 hrs. ( providing this is the flow and flash point temp of the PLA.) move to 1450.f for a complete burn out of the can. ( the can will be completely chalk white when burnout is complete yours is not.) Although your getting a complete pour your porosity will significantly be reduced if you have your molten metal ready before you pull the can from the kiln which should be slightly warmer than the melt temp of the metal your working with providing your using a cold vacuum system. the vacuum will rapidly cool the can causing cracking and fins etc. so priming I would assume is when you set the can on the pad and you have achieved a full vacuum. Timing is particularly important with these metals. too cold porosity again then you have to make sure the sprue is thick enough as well as the metal cooling in thin to thick areas are going to yield more metal flow disruption and cooling. ( heavy areas cool slower than the thin ones and usually you find problems at or along the junction of those areas)
    if you pour the metal into the can as soon as the can is set and vacuum is achieved it will be much better no matter what metal you use. It is a balance to keep the metal flowing into the can without solidifying on the way in causing the “skin” on the outside of the casting. A perfect casting in any metal yields a semi shinny surface after pickling to remove investment,

    • bowman says:

      This is true, and these are all things in the 100+ pages of documentation. However, the goal was to offer a primer… something that kept the concepts concise and approachable. Approximately 600 words concise, your comment would have taken half the article. Yes, the burnout can idle between 600-800 degrees F for hours until the piece has drained, but that is dependent on flow rates, venting, sprue design, and initial weight of the piece, (not for a beginner). The kiln schedule is a baseline, as people gain understanding and confidence they are free to modify their burnouts to accommodate their pieces… the “open” part of open3DP. You don’t want to bring the investment past 1250F “Satin Cast 20 tends” to degrade and get fissures/flashing with the PLA expansion rates. I understand 1450F is past the point of instant ignition and will ensure a complete burnout, however that is weighed against the detail and integrity of the PLA part being cast. The porosity in the aluminum is because it is recycled scrap 6061 aluminum, these artifacts were not present in fresh cast pieces of ingot aluminum. Thank you for your enthusiasm, criticism, and dialogue.

  3. […] previous article “Rapid Manufacture: Iteration and Industry” documented how it is now simple and cost-effective for any business/start-up to manufacture […]

  4. Jb says:

    I plan on printing in PVA soon and casting in aluminum. I believe it will burn out more easily on small parts than PLA. Small parts didn’t burn out well for me before the plaster degraded. My walls were less than 1/8 which I understand is the minimum for plaster of Paris (even with sand mixed in).

    Have you tried using PVA instead of PLA?
    I would love to hear about that.
    Jbc

    • bowman says:

      PVA should burn out fine the melting points are similar to PLA. Remember that hydroxyl (alcohol) groups make the polymer polarized and hence more water soluble. The prints may partially dissolve away in the investment, or worse repolymerize with the catalyst in the plaster [causing cauliflowering]. The easiest way to test this is to pour an open faced mold over a printed part that is drafted or, something can easily be removed from mold, to check if they bond. If so it is likely that the PVA will have to be sealed or desiccated out before the investing process.

  5. […] the water evaporates from the investment [see investment instructions (above) for burnout timing or burnout schedule here.] Normally a kiln will have a huge lag as it tries to get all that thermal mass up to temperature. […]

  6. aonemarine says:

    Looks like you guys are still guessing….follow the mfg burn out cycle and learn what model s can be cast without issue. There are limits on the pla casting when it comes to certain geometry due to the high expansion rates fracturing the investment. Blind holes are the least forgiving.

    • bowman says:

      Guessing is not necessary, the data is all there. Most things that require casting are usually too complex to try and explain individually. Each scenario has it’s own set of fluid dynamics. There is a more in-depth post if you are still having trouble with the process.

      Found here: http://open3dp.me.washington.e.....ing-guide/

      It helps to pack flasks more tightly. This prevents any one singular vector from creating shear forces that cause fractures to occur. If the PLA is breaking the investment, either the investment is mixed poorly, or the burnout is being rushed. If the flask is pushed through [200F… 100C] water’s phase change [water to steam] in can easily fracture the flask.Then the burnout is simply limited by the heat transfer from the outside of the flask to the core. If the PLA can push that hard than the infill may be too dense, infill should be 3-8%. Otherwise, try a slower ramp up from 100-300F.

      Small blind holes are generally post process, they are drilled out after casting to maintain appropriate tolerances. Any blind hole under 5mm tends to trap air bubbles. Parts should be designed with the intent to scribe/mark holes to be drilled in the part during post production tolerancing.

      • aonemarine says:

        Im not having any trouble, quite the contrary I get excellent results.
        http://youtu.be/20iYekoQUUA
        Where I do run into trouble is when it comes to printed text (small blind holes) like the center of an “o” breaking the investment loose
        Good to hear from you bowman. Ive been following you on youtube. I will read the extended cut version of the article. It will be interesting to compare it to my findings.

        • bowman says:

          Yes raised texts are a nightmare when it comes to fluid dynamics, one of the most common places where shear forces are a constant. The best way to resolve that is a 1-10% draft angle. We had a good test case where it was troublesome to do thin wall draft-less angles in excess of 4.5-5mm heights when thickness is less then 1.6mm [2% draft angle resolved the issue]. Those parameters can be trouble shot with these test parts:

          http://www.thingiverse.com/thing:525791

          If the letters are still troublesome then a draft angle may be necessary. The data did show consistently good results on dimensions smaller than .8mm for aluminum, and those are 2% draft angle edge cases. There are more posts coming with additional data, but one can only work… and document so quickly.

          • aonemarine says:

            Looking forward to seeing your testing of blind holes. I’m well aware of increasing the draft angle as well as filliting the edges. When I was testing different investments it was with using blind hole models. I would cast about 20 of the same item in a flask all with the same depth and draft angle to see how many would come out. Then I would change the draft and cast 20 more. this test was repeated so many times I don’t ever want to do it again. There was noticeable improvement by increasing the draft, but by no means a fix all.
            Blind holes are trickey buggers and have been very difficult to overcome unless you remove the compressive forces from one plane. But then the geometry of the hole as modeled has changed so you might as well just create the blind hole in a finishing operation. It would be interesting to create a model with different diameters and depths of blind holes and alternate the draft angle on different models to get a better understanding of what can be cast repeatably. Quite honestly I have grown tired of blind holes and choose to avoid them when I can.
            I have been casting with the lost pla method since May 2013 and casting metal for over 8 years now. If I create a model file for you to test blind holes would you be willing to cast it.and document it?

  7. aonemarine says:

    Bowman, I read the more in depth post and looked at the casting defects. I had the same type of defects when I started playing with the lost pla casting and have resolved those issues.
    I have been persuing lost pla casting for quite some time and have a great deal of experience with the process. Drop me a line sometime, would love to chat.

Leave a Reply