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.
I think that applies to any use of 3D patterns, if you have layer marks on the surface then they are likely to be carried over onto the casting.

In fact just about any form of pattern with surface defects can carry over be it laminations of poorly sanded plywood, sunken filler used to fill holes etc. You only get out a sgood as you put in.
I recall thinking about using a recess in a 3D printed pattern.

For example, with head that has vertical and horizontal fins, hollow out the center of the pattern, and insert a 3D printed piece, ram the mold, remove the center insert, then heat the rest to remove the plastic from the fin area, both vertically and horizontally.

Last edited:
The terminology is a bit confusing to me, since I have only done sand mold castings.

This guy discusses "investment castings" vs "ceramic shell".

I think using either method, you can still use the "lost PLA" method, ie: the PLA pattern is burned out in a kiln.

He mentions that the ceramic shell method is better for larger castings, since you would have to use a lot of investment material with a large pattern and flask.

Cracking during burnout seems to be a major drawback with the investment method.

I have seen a number of ceramic shells crack too, using the lost-wax method.

Having watched this investment video, I think I would lean more towards the ceramic shell lost-PLA method, if I were to try this type of casting.

I am trying to think of shapes on the various engines I have modeled in 3D that would force me to use a lost-PLA process, and I can't really think of a part that I can't cast using bound sand.

Pretty interesting video.
He tries several different investment materials.

For lost wax or lost PLA castings, the major advantage of investment and/or ceramic shell methods seems to be casting a large number of small parts in a tree arrangement, which would be difficult to achieve using sand molds.

An example is in this video.

I really don't see much if any advantage to using the lost-wax method for making most model engine castings.

The lost-PLA process I think lends itself more to model engine castings, as we have seen in some recent examples posted on this forum.

Last edited:
I considered purchasing one of these small electric furnaces, but a buddy of mine got one, and he said crucible life was a problem.

This video seems to reinforce what he told me.

Here is myfordboy illustrating the lost-PLA process, using investment.

The complexity of the part is high, and thus pulling it from a sand mold would be difficult and/or perhaps impossible.

So he makes a good casting from a complex pattern, but the problem remains with the 3D print lines.

As I mentioned above, I have seen one individual fill the PLA pattern using spray paint, and the paint successfully burned out, so he did not have the 3D print lines.

This is not really a part you would see on a model engine though, and so it would be difficult to justify the added complexity of the lost-PLA process if the part could just as easily be cast using greensand or bound sand.

I feel that if I go to the trouble of filling a pattern, using paint or whatever, the last thing I want to do is use it a single time and burn it out using the lost-PLA method.
The lack of having a reusable pattern is what I consider a big downside to the lost-PLA method.

Last edited:
I see a lot of folks using sodium silicate for cores, but it can also be used to make the entire mold (bound sand mold).

Here is an example of that.
It is important to use an exact amount of sodium silicate per pound of sand, else it will be very difficult to break the mold down after pouring the casting.

And sodium silicate will crumble if over-gassed with CO2.
I generally use 5 seconds maximum if I use CO2.

Sodium silicate molds can also be hardened using a catalyst, which eliminates the need for CO2.

I made a few molds using sodium silicate bound sand, before I switched to resin-bound molds.
I think the resin-bound molds are easier to work with than sodium silicate bound molds, but sodium silicate bound molds are a viable option, and they don't have the fumes that the resin molds has.

And I think ceramic mold coat would work with sodium silicate molds, but I have not tried that combination.

I have seen a large full-sized propeller cast in a sodium silicate mold successfully, and some sort of full-sized pump housing.

I think sodium silicate bound molds are under utilized in the model engine casting world.

Here is an open-faced pour.
I don't see that very often, but it is a viable option in some cases.

Don't stir an aluminum melt, and don't add salt to it either.

Yes the "foam" type PLA has less material to burn out, someone posted about it here a month or so ago.

Pat I don't see why you say trees of castings cant be done with sand moulding. It is really no different to using a match board with multiple parts on it with all the runners and gates. This is a very commonly used method as it saves time having to mould each individual part in a separate flask, reduces total amount of metal to be melted as you don't have as many risers etc.

I've even made patterns that consist of several items "treed" together so they are quicker for the foundry to cast and less risk of loosing individual small patterns. This is probably a better option for the home foundry as they are not going to be into production quantities that match boards suit.


  • 20240526_073221[1].jpg
    2.5 MB · Views: 0
  • flange stick.JPG
    flange stick.JPG
    64.7 KB · Views: 0
  • 20220227_144040.jpg
    2.3 MB · Views: 0
Your engine parts probably don't need to be done with lost PLA as they were designed from the start to be cast in sand with cores.

But you could eliminate the need for quite a few cores simply by doing the same items with lost PLA. may take a little longer to do the shell but no corebox or core to make so overall may be quicker. Think of that Monitor crankcase, you would not need to core that if done as lost PLA and as there is no separate core you won't get flash where the iron tries to seep between pattern and core around cut outs like where the crank bearing openings are. Also eliminates visible parting lines so less clean up of the casting afterwards again saving time over using a split mould.

Same would apply to that bottle engine, no need to worry about core sand a nice flash/line free surface around th etwo big side cut outs. Ditto the Galloway bed casting a lot easier to mould as you don't have to worry about the overhanging bearing housings preventing a straight pull.

Where investment or lost pattern is at it's best is on smaller complex items, think of engines like some of the old Breisch ones where you got the basic iron parts then you could also buy an additional set of the smaller items in bronze, these would suit the lost wax process (not sure what they used) but I've certainly made engines where some of the castings in the kit were investment cast. Simply use what method suits the individual part best rather than try and stick with one thaat may only be best for a certain type of pattern.

Best thing to do it try it out, spend the time casting rather than trawling the net for videos 😄
Last edited by a moderator:
Regarding the time for sand vs burn out the whole job should be considered not just the actual time to get a mould.

If using a printed pattern and green or bound sand the pattern needs to be fairly robust which means a thick wall and reasonably high amount of fill. This will add considerably to the print time as well as the cost of filament used.

On the other hand a print intended for burn out needs little strength as it only needs coating so can have a thin wall and minimal fill. So print time is a lot less and you could print several in the same time as you could a more solid pattern. Also easier and quicker to burn out less material as well as the saving in material cost.

here is an example of an engine I did the CAD for recently. This is a shot of an unmachined original casting, you can see from the parting lines that it was not a simple top and bottom to the flask and an easy pull mainly due to the crankshaft boss sticking out both sides of the vertical support. It is also hollowed out underneath and that bit ontop of the bearing housing is cast hollow to act as an oil pot.

ideal 5.JPG

To keep costs down it was decided to leave the bearing boss off and simply JBWeld in a piece of steel for the model

ideal 6.JPG

This would have been an ideal candidate for burning out the printed material as the boss could have been included as well as having zero draft where the cylinder bolts to the bed casting, rough hole for cylinder liner would not need a core.

1/3rd scale complete, 1/2 scale ready to cast

ideal 7.JPG
Pat I don't see why you say trees of castings cant be done with sand moulding.

Now that you mention it, I do recall seeing many of the original Cretors patterns at Bob Pearson's shop, and they were flat tree'ed.

Photos attached.
One of the photos is a lost-wax mold that Bob had made.

One thing that surprised me is that Cretors did not use split patterns for anything.
All Cretors original patterns are one-piece.
I presume Cretors used a follower where required, on things like the No.06 frame.



the Galloway bed casting a lot easier to mould as you don't have to worry about the overhanging bearing housings preventing a straight pull

If the overhanging bearing housings have a draft on them, they do not prevent a straight pull.

The Ball Hopper Monitor crankcase would be a good place to use a "lost" something process, and the bottle engine frame.

All the old engines used cores though, and so while the lost method may be convenient, it can always be done with cores.

The Soule Speedy Twin is about the trickiest engine I have tried to make patterns for, and for a small scale Speedy Twin, the lost method would be very useful.
The Soule Museum guys want a 1/4 scale Speedy Twin, and I told them I could not sand cast that, but I could use lost PLA for that application.
The passages in the top of a Speedy Twin are extremely complex, and would be very difficult to cast on a small scale.

The space between the cylinders is rather small.

This is an incomplete 3D print for the Speedy Twin, with dimensions taken from a full-sized engine.
I almost have that 3D model complete.

This is probably the best example of where a lost PLA process would help a lot.

Best thing to do it try it out, spend the time casting rather than trawling the net for videos

I am on vacation at the moment, and so many miles away from the shop, as if I had time to get out to the shop if I were there.
Retirement is coming one day, but I am not sure exactly what day that is.

If using a printed pattern and green or bound sand the pattern needs to be fairly robust which means a thick wall and reasonably high amount of fill. This will add considerably to the print time as well as the cost of filament used.

If you intend to make a permanent pattern from a PLA print, you can use a minimum infill, double shrinkage, and make the pattern just strong enough to make one mold, then cast a permanent pattern in PLA.

Somebody did not do their parting line anything like I would have done it.

A vertical parting line would work fine in my humble hobby-person opinion, with an core insert or two.

Last edited:
Can't see how a vertical line would have worked either across or along the part. Though I have seen more of the item than you.

If you google Hardy and Padmore and then click images you will see from their products they were a well established foundry and engine maker, and I expect they knew a lot more than you.
Last edited:

Latest posts