Semi-Lost-PLA Casting Method

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

GreenTwin

Well-Known Member
Staff member
HMEM Supporting Member
Global Moderator
Joined
Jul 2, 2021
Messages
3,893
Reaction score
1,548
Location
MidSouth, USA
I recall some conversations not too long ago about a semi-lost-PLA method, but I forget if anyone actually tried the method, or whether it was successful.

Our member Foketry is doing some most impressive lost-PLA work at this thread; see Post #22, etc.

https://www.homemodelenginemachinis...s-as-a-draw-tech-design-upscaled.36071/page-2

The last time I tried to cast a similar part was with a windmill gear, and I got very close to making an accurate windmill gear in gray iron, but could not get a clean enough mold to make as accurate a casting as I wanted.

The gear pattern had no draft angle on it, and sometimes with bound sand, you can get away with not having any draft angle on a pattern.

A few photos of the windmill gear attempt are below.
The gear pattern was 3D printed by the owner of the windmill, not by me.
It was a really nice 3D print, and very smooth.

The reason I attempted to cast this gear is not that I needed a gear (the windmill guy needed two gears for an antique windmill), but rather to see just how far I could go with a pattern that had zero draft angle.


rImg_2488.jpg
rImg_2494.jpg
rImg_2495.jpg
 
I know from visiting the Soule museum in Mississippi that casting gears in the old days was a common thing, and that museum has a large number of all sorts of gear patterns.

I am not sure exactly how they did it, but it is obvious that they make molds for gears, and then cast gears.


fImg_5017.jpg
rImg_4988.jpg
rImg_4989.jpg
rImg_4992.jpg
rImg_4997.jpg
rImg_4998.jpg
rImg_5006.jpg
rImg_5007.jpg
rImg_5008.jpg
rImg_5009.jpg
rImg_5010.jpg
rImg_5074.jpg
 
I tried a multi-piece bound mold with the gear.

Despite using a commercial pattern coating product, the gear pattern was damaging the mold when I attempted to extract it.

I experimented with several flask configurations, and was not really sure where to put the gate(s), riser(s), runners, spin trap, etc.

r20190420_212229.jpg
r20190420_221956.jpg
r20190420_224740.jpg
rrImg_3933.jpg
 
My intent was to cast gears that could be cast and then used without machining, just by polishing them.
With bound sand and ceramic mold coating, you can get a glassy surface on a casting, without the typical sand grit texture, such as seen on this iron casting that just came out of the mold, with nothing done to it as far as surface prep/cleanup.

rImg_7743.jpg
 
Last edited:
Looking at that ad my first thought was "That's cool, that'd be only half the stress on each gear set."

But then again it's also twice as many parts that can fail. How did this idea work out for the company?
 
After breaking the first mold during pattern extraction, out of desperation I tried greasing the pattern.

I was finally able to extract the pattern, but the grease in the mold was not a good thing, and it threw off the accuracy of the casting.

I decided to concede defeat on trying to cast an accurate gear at this point.

There were some other folks who did cast gears for this project, but they either had draft angle, or where not accurate enough to be usable without machining.

.

RImg_3911.jpg
RImg_3915.jpg
RImg_3917.jpg
RImg_3923.jpg
rImg_3929.jpg
rImg_3933.jpg
rImg_3935.jpg
rImg_3937.jpg
rImg_3939.jpg
rImg_3943.jpg
rImg_3952.jpg
 
Looking at that ad my first thought was "That's cool, that'd be only half the stress on each gear set."

But then again it's also twice as many parts that can fail. How did this idea work out for the company?

I have to admit I don't really understand exactly how the various windmill types work, other than at some point they must raise the pump shaft up and down.

I do recall some manufacturers discussing efficiency, and the ease of pumping with low wind.

Baker (the same folks that made the Baker Ball Hopper Monitor) was a major windmill manufacturer, and I read of description of their very elaborate display that was at one of the big world machinery expositions.

.
 
The beauty of attempting to cast a part like this is that even if you fail, you can still learn a lot, and perhaps get closer to success.

Here are some of the multi-piece mold parts.

.
rImg_3957.jpg
rImg_3958.jpg
rImg_3959.jpg
rImg_3960.jpg
rImg_3961.jpg
rImg_3962.jpg
rImg_3966.jpg
rImg_3967.jpg
rImg_3972.jpg
rImg_3973.jpg
rImg_3974.jpg
rImg_3976.jpg
rImg_3977.jpg
rImg_3978.jpg
 
I gave up on trying to cast the gear, but I did get some nice parts cast later that day, in gray iron.

These castings show the surface finish right out of the mold, with no cleanup, and this is due to the ceramic mold coat that is sprayed into the interior of the molds.

rImg_5766.jpg
r2Img_5787.jpg
rImg_5780.jpg
rImg_5783.jpg
rImg_5787.jpg
rImg_5808.jpg
rImg_5813.jpg
 
Getting back to the title of this thread, which is "Semi-Lost-PLA Casting Method", I have been pondering how to come up with a "lost-PLA" method that does not require the investment coating and oven burnout.

For bound sand castings, the method is as follows (for a 3D printed pattern):

1. 3D print the pattern.
2. Mix the sand with the binder, and ram the mold.
3. Extract the pattern from the sand.
4. Pour the mold.
5. Pattern is reusable multiple times.

This entire process can occur in as little as a hour or two, and so there are very few steps, and you can actually make several rapid-prototype attempts in a single day.


For the lost-PLA process (as I understand it), we have the following:

1. 3D print the pattern.
2. Dip the pattern in investment multiple times (not sure exactly how many, or if there is a dry time between each).
3. Burn out the PLA pattern in a kiln.
4. Pour the mold.
5. Pattern is lost during this casting process.


The pro's of the bound sand method are that it is a relatively fast and simple process, with few steps.
The con's of the bound sand method is that it may not work with intricate parts, especially if a draft angle cannot be used, or if the shape of the part prohibits it from being withdrawn from the sand mold.

The pro's of the lost-PLA method are that you don't need to worry about molding sand, mulling the sand, etc.

The con's of the lost-PLA method are that you have to purchase the investment product, and it is not resuable (bound sand is also not easily reusable); sometimes the investment shell cracks during burnout, and sometimes you don't get a completely clean burnout.

.
 
So the discussion was about how to come up with a hybrid process that could push the bound sand method more into the realm of lost-PLA, without having to do the kiln burnout or the investment process.

When alcohol-based ceramic mold coat is sprayed onto the inside of a bound mold, it is burned off.

Generally the bound sand mold is burned off before the ceramic mold coat is applied, and so it is known that bound sand molds will withstand at least a light flaming with a soft propane flame.

You can see the mold coat burn-off process in this photo.



rImg_5749.jpg
rImg_5751.jpg
 
One example of just how far you can go with bound sand molds is Jeff Henise's racing motorcycle cylinder castings in 356 aluminum alloy.

https://www.homemodelenginemachinis...se-casting-motorcycle-racing-cylinders.35502/

Jeff used a draft angle on his fins, and was able to successfully extract his patterns from bound sand.
There was some rather complex core work also used in the molding process.

So you can really do a lot with bound sand molds, especially if a little draft angle is used.

Photos by Jeff Henise:
At first glance, one would generally assume that these cylinders must have been created using the investment process.


103882-92673-104.jpg
103881-92674-105.jpg
103875-92680-119.jpg
103872-92666-Image47.jpg
103870-92668-101.jpg
103869-92669-102.jpg
103866-92672-110.jpg
 
Pat. Those pictures of the sand cracked into four around the gear should give you an idea of what can be done. Split the flask into 4 or six pieces and then ram the sand like slices of a pie. You can then pull the sand away sideways using the angle of the teeth as your draft rather than try to pull the pattern out vertically.

The other method would be to make the pattern from multiple parts that can individually be moved towards the middle and then lifted out

Third method is to make a core box for say just 3 negative tooth saces at a time and then make several cores which can be assembled to make the mould There are some videos of teh guts in india using a similar menthod to cast big gears on Youtube
 
My thoughts for 3D printed patterns with no draft angle are to put the bound mold and imbedded pattern in the oven, and heat them to the point where the PLA material is soft and flexible, but not fluid yet.

If the PLA material is soft, and using a minimal amount of infill inside the 3D printed pattern, then when heated, the PLA pattern should collapse to some extent, and thus allow the pattern to be withdrawn from the sand mold, even with no draft angle.

Bound sand molds can also be made in multiple pieces, and those pieces can be adhered together using mold adhesive, so that could further extend the possibilities of what could be molded.

Some experiments that I need to try are heating a bound mold with an imbedded PLA pattern, to find out just how much heat the mold will take without deteriorating, and to find out the exact temperature to make PLA soft but not fluid.

Perhaps this method would not go as far as what can be achieved with the investment method, but I think many would be surprised at how complex a part could be cast with no draft angle, using bound sand, and heating the PLA until soft to withdraw it.

.
 
Pat. Those pictures of the sand cracked into four around the gear should give you an idea of what can be done. Split the flask into 4 or six pieces and then ram the sand like slices of a pie. You can then pull the sand away sideways using the angle of the teeth as your draft rather than try to pull the pattern out vertically.

The other method would be to make the pattern from multiple parts that can individually be moved towards the middle and then lifted out

Third method is to make a core box for say just 3 negative tooth saces at a time and then make several cores which can be assembled to make the mould There are some videos of teh guts in india using a similar menthod to cast big gears on Youtube

I want to try heating the bound mold to soften the PLA pattern, and see if that would make for an easy withdraw.
This would be a simple solution, and the methods you mention are not necessarily simple or as accurate.

I am sure I have asked before, but has anyone tried heating a PLA pattern to soften it before extracting it from a bound sand mold?

.
 
How would you heat the pattern once it is in the set sand? Soulds like you would risk the pattern distorting and may have to print again if it did not pull out cleanly.

Pics of the third method I suggested, Th ecorebox could be 3D printed rather than this CNC cut one. Finished gear is about 6ft diameter

Corebox

gear 1.JPG


Core in the box

gear 2.JPG


Mould made with large plywood pattern having teh individual cores set in position, not mouldwash on teeth





gear 3.JPG


Finished iron casting just after blasting

gear 4.JPG
 
How would you heat the pattern once it is in the set sand? Soulds like you would risk the pattern distorting and may have to print again if it did not pull out cleanly.

One idea is to put the bound sand mold with pattern in the oven, and heat it to a point where the PLA begins to collapse, but not become fluid.
Yes, there is the risk of the mold having to be made again if you don't get the pattern completely removed, but if I can almost remove a 3D printed pattern like the windmill gear from bound sand without heat, then I think with a little heat, it would be no problem at all to remove something like the windmill gear pattern without damaging the sand mold.

There is a possibility of a heat gun being used, but this is more risky, since you would not get the entire pattern hot and soft.

The gear casting you illustrate is a fantastic process, and that is exactly how they use to mold large gears and flywheels back in the 1800's, but the intent is to come up with a process that is only slightly more complex than a typical bound sand mold process, but without requiring draft angle.

So the intent is achieve lost-PLA ceramic shell results using a simple bound sand mold with PLA pattern; ie: no burnout, no ceramic shell, no shell cracking, and a very quick molding process that is ready to pour in under an hour.

There is no need to break the pattern/corebox into multiple parts unless there is some unavoidable reason that forces that situation.

So "semi-lost PLA" process, ie: soften the pattern (but not melt it), to allow the use of a pattern with no draft angle and perhaps some thin parts like cylinder head fins.

.
 
I can see it's uses for something like a gear where you want the teeth square with no draft.

Cylinder fins can be printed with draft even if only 1 degree. but even then you are limiting yourself to fins in the direction you want to pull the pattern. Not going to work on something like that knucklehead where fins are vertical on the top but horizontal around the edge.
 
Not going to work on something like that knucklehead where fins are vertical on the top but horizontal around the edge.

That is a good point, and perhaps when your multi-piece pattern would come into play.

At some point a multi-part mold may get more complex than just using the investment method.

I am basically trying to think outside the box, so to speak, and perhaps find some easier, faster, simpler methods for making the more complex engine castings without having to resort to the investment/burnout process.

One of the drawbacks of the lost-PLA process is that any rough surface finish gets mirrored over into the castings.
I have seen one lost-PLA casting method where the pattern was painted to conceal the 3D printed lines, and the paint seemed to burn out during the burnout process, thus leaving a very smooth casting using the lost-PLA process.

Example below:

 
Back
Top