Home Foundry

[GreenTwin]

I have followed an individual for several years who was the first to my knowledge to cast a V-8 engine all in gray iron.

If you can make the cores for a V-8 manifold, and successfully cast one of those in iron, chances are you are pretty good with the iron stuff.
Everything has to be exactly right to fill intricate molds, but as he has shown, it can be done.
It took he a while to figure it out.

.

 

[dazz]

Here is what I am aiming to cast. None of that simple easy stuff. This is a compound engine top casing for a Yorkshire Patent Steam Wagon. The prototype is only one of 3 remaining. It is located in Invercargill, New Zealand. It is an ugly duckling but quite unique.

The plan is to build a 1/3 scale miniature. This CAD drawing is produced from the factory drawings held in an archive in Leeds UK. The green flange is the beginning of the centre casing. The original full size casting was only 5/16" thick. Too thin to cast to scale.

I am not going to start learning casting on this piece. There are plenty of other less complex parts to make first.

 

[GreenTwin]

There is nothing like a good challenge as I say, and that crankcase will take some thought.

Very nice 3D work there for sure, and a great unique (unique to me at least) engine.

There are several approaches to intricate castings such as the one you illustrate.
One is the "lost foam" approach used by industry to cast engine blocks and other objects such as pump housings etc. that don't lend themselves to a traditional pattern/sand mold method.
I have studied the lost foam method, but without the exact foam material, it would seem that the results are substandard.
And another problem with lost foam is the need to make a permanent/semi-permanent die in which to cast the foam.

Another approach is the "lost PLA" method, and this method actually works if you use a 3D printer filament that burns out cleanly.
The lost PLA lends itself to smaller parts though, mainly due to 3D printer size limitations.

Lost wax is similar to the other "lost xxx" methods, but again, it required making a rather intricate die in which to inject the wax.
I have not seen lost wax used on a large scale.

There were some intricate objects cast back in the old days such as lathe beds, or even steam engine crankcases, and these used greensand molds.
One engine in particular that I have studied extensively is the Soule Speedy Twin.
I live a few hours away from the factory that manufactured the Speedy Twin, and that factory is still intact, and most of the patterns and core boxes are still at the foundry that is located in a building adjacent to the factory.

The secret of the Speedy Twin engine casting was the ability to make some very intricate cores using I believe linseed oil and sand, which was then baked to create durable hard cores for the multitude of passages that are located in the top of a Speedy Twin engine.
Baked linseed oil cores were the early equivalent of the modern bound sand cores, and the results were very impressive for the early 1900 period.

Luckily the original Speedy Twin crankcase corebox is still at the factory, and I have studied it in great detail.

For complex castings such as steam engine crankcases, "retracts" are your friend.

The second secret to successful Speedy Twin crankcase castings is the extensive use of "chaplets".
Chaplets are small pieces of steel that are used to hold the cores in an exact position during the mold fill.
The chaplets become part of the castings, and are generally totally enclosed by the molten metal, or nearly totally invisible after casting a part.

Cores have a tendency to float or shift position while the mold is being filled, and the lifting force on a core is very significant, especially with iron.
Cores also tend to shift due to the stream of iron striking them during the mold fill, since iron is very dense.

 

[GreenTwin]

The good part about copying an existing engine design that was actually manufactured is that you know that someone long ago was able to cast the part using greensand and perhaps baked cores.

For new designs, one has to figure out if the part can be cast without resorting to the lost foam or some similar method.

Here is the incomplete 3D model for the Speedy Twin.
The Speedy Twin is an extremely compact and powerful engine that was designed to operate sawmill carriages.
For many years, its performance (in measured output of sawn timber) was unmatchable by any other engine or even by electric motors.

The Speedy Twin is a unique engine in that it is fully reversible, and yet it only has two eccentrics (one for each cylinder), and it does not have a valvegear reversing mechanism such as a Stephenson's link.

The Speedy Twin has a D-valve for each cylinder, and the pressure is reversed on the D-valves when the engine is reversed, and yet the D-valves do not lift off their seats (I defy anyone to explain how this is possible; it took me a year to figure out how it can be done).

 

[GreenTwin]

The good part about the Speedy Twin is that it seems to be a unique and highly functional "designed for a specific task" type engine, similar to the Dake engine.

The bad part about a Speedy Twin, if you intend to build one, is that the engine was way ahead of its time, and is an extremely advanced design, both in function and in the foundry work involved. I have never seen a steam engine design like the Speedy Twin, and I think it is a one-of-a-kind design.

Luckily I ran across a disassembled Speedy Twin for sale a few years ago, and so I have all the parts for the engine; otherwise I don't think I would have ever figured out how it functions.

The Soule factory (now a museum) displayed one of the patent drawings for the Speedy Twin, with highlights in red showing how the steam flowed in the maze of passages. I looked at that diagram for a year, and then told the folks at the museum that their diagram was incorrect.
They did not believe me at first, but I proved to them how the Speedy twin steam actually flowed in the passages on a working engine.

Here is a good video of a Speedy Twin being reversed.
The efficiency of the Speedy Twin design was that it could reverse the sawmill carriage very quickly, and since a steam engine produces 100% torque at zero rpm, the speed at which a Speedy Twin could return the carriage to the opposite position was unmatchable by any other mechanism available at the time.

Steam enters the steam chest on the top of a Speedy Twin, where it is controlled by the reversing valve, which is a single D-valve arrangement.
The exhaust is the hole on the right of the steam chest, and in the video, there is no exhaust pipe attached to the engine.

 

[GreenTwin]

Here is a 3D print I made for the Soule museum folks.
It is not a complete design, but rather how far I had progressed on the engine at the time.
More of a table ornament type print, but it could be used to build an engine if the lost wax/PLA method was used.
It is an exact (not quite finished though) scale model of a Soule Speedy Twin.

There is speculation about how large or small of a scale the Speedy Twin could be built.
The museum would like a 1/4 scale model.
I am sure some out there could build a Speedy Twin at that scale, but then some models use microscopes to built model engine parts.

I am more of a big-scale builder, since I have trouble machining on a watchmaker's scale (or any small scale), and have further trouble seeing small parts.

 

[GreenTwin]

Here is what I am aiming to cast. None of that simple easy stuff. This is a compound engine top casing for a Yorkshire Patent Steam Wagon. The prototype is only one of 3 remaining. It is located in Invercargill, New Zealand. It is an ugly duckling but quite unique.

The plan is to build a 1/3 scale miniature. This CAD drawing is produced from the factory drawings held in an archive in Leeds UK. The green flange is the beginning of the centre casing. The original full size casting was only 5/16" thick. Too thin to cast to scale.

I am not going to start learning casting on this piece. There are plenty of other less complex parts to make first.

So back to the crankcase on this engine.

One thing I learned by trial and error is the art of transforming a 3D model into a pattern.

If a sand mold will be used, generally speaking (there are no absolutes in foundry work) there needs to be draft angle on all the surfaces (sometimes I do not use draft angle on parts, and there is an art to doing that successfully).
The draft angle allows the part to be withdrawn from the sand mold without damaging the mold.

The second consideration for creating a pattern from a 3D model is machining allowances.
The surfaces to be machined must have an extra layer of material added to them in the 3D model, so that when those surfaces are machined, they will have the correct dimensions.

And the correct shrinkage factor needs to be considered for any patterns that are 3D printed, ie: the pattern must be larger than the final cast part, since metal shrinks as it cools. For gray iron, I think I generally use a shrinkage of 0.015 uniformly, which generally gets me in the ballpark and creates usable castings of the correct dimensions.

I suspect you will need several retracts, which are parts of the pattern that can be separated, and the retracts are generally doweled into the correct position, with the dowels allowing the pieces to be separated using a sliding motion.

I would definitely use bound sand for this piece (I use bound sand for everything including molds and cores).

.

 

[GreenTwin]

For my 3D models that are converted into patterns, I sometimes use a coreprint, which is what was used on the Speedy Twin.
The coreprints are protrusions in strategic locations that allow for support of the cores, such as the bore cores.

Scaled down engine parts do tend to get too thin, and I often add material on the inside of the pattern, so that the outside of the casting remains to scale.

You can also glue multiple cores together, and that is yet another trick to successfully casting an intricate engine.
The core glue is a ceramic-type material generally what will withstand heat long enough to let the iron solidify.

I first thoughts on casting that crankcase would be to face the open window upwards, with the crosshead guides horizontal.
I would fill it from the bottom using two long knife gates.

Sometimes you have to cast a part that contains a window as a solid, and then machine out the window area, in order to get a complete mold fill.
I suspect you may have to do that to get that very thin part to fill.

Bore cores must be proportionally smaller than the final machined bore size, which is a way of adding machining allowance.

I don't try to cast bolt holes, although I have seen bolt holes accurately cast using the lost PLA/wax method.
It is easy enough to drill and tap a bolt hole, and I see no need to try and cast bolt holes.

.

 

[GreenTwin]

For your pattern, I would probably section the crankcase through the crosshead guides, and anything protruding 90 degrees to that plan would have to be a retract of some type.

You could have one large coreprint coming out the bottom of the crankcase, and then use a modified 3D printed pattern as a corebox to make a core that captures all the interior details.

.

 

[dazz]

Hi
The current 3D CAD drawings are straight copies of the original 2D drawings. Once they are completed, I then need to scale the drawings and modify them to produce 3D drawings suitable for making patterns. This includes machining allowances, pattern draft, filling in bolt holes, shrinkage, cores etc.

The factory drawings do not include any draft. Changing the design/pattern to include draft would have a lot of downstream side effects. It will be easier to make a complex, multipart pattern, than change the design to include draft. Fortunately I already have a 3D printer.

I plan to use Sodium Silicate bound sand because it is obtainable.

It will be a long time before I am ready to attempt to cast this. The bottom case, and maybe the centre case, may be easier to fabricate rather than cast.

 

[GreenTwin]

The factory drawings do not include any draft.

I think much/most of the required draft angle is inherent in the design; ie: most of the surfaces are either sloped, curved, or circular in shape, and so I don't see the need for much if any draft angle that needs to be added to the model.

The bottom case, and maybe the centre case, may be easier to fabricate rather than cast.

But much more interesting to cast because then you will know how to cast a lot of engines and other things.

I think I would split the crankcase into four pieces.
I would tape the interior of the patterns with blue painter's tape, ram the cope with two taped pattern pieces, ram the drag with two taped pieces, break the patterns slightly from the molds but leave the patterns in the molds, join the mold halves, and then ram the interior to make the core.

To remove the core, separate the two mold halves, and then remove 1/4 of the pattern at a time.

Piece of cake.

Sodium silicate bound sand will work I think as well as resin-bound sand.
Use plenty of wax or mold release on the patterns.
Sodium silicate seems to try to stick more to the patterns than resin-bound sand.

Here is an example of someone casting a large intricate engine part.
I would not have done it the way he did it (I recommended he make the entire mold all at once instead of in multiple parts).
Still an impressive casting, and so close to being a good casting.

https://oldschool.co.nz/index.php?/topic/53938-bugatti-t5759-engine-project/page/5/

 

[dazz]

Hi
All good advice.

I foresee the problem with Sodium Silicate is that it melts at 1088 deg C, that's below iron mp. So I would expect the sand in contact with the iron to become separated from the mold and embedded in the iron.

I am wondering about doing a home-brew ceramic mold coating with something like Zirconium ceramic powder with a binder. I can buy raw powder from China. The problem being that I have no idea what chemistry is used in the commercial products. There is absolutely no danger of me being able to by a proper foundry mold coating. I may be over-thinking this problem.

 

[GreenTwin]

I don't think I have ever tried using sodium silicate with iron.
Perhaps with a core or something; I can't really recall.

One of the reasons I switched from sodium silicate (when I was doing aluminum castings) to resin binder was that I only knew about the CO2 method of setting the SS bound sand. Sodium silicate bound sand is very sensitive to SS percentage (3% max worked best for me), and very easy to over-gas, as well as undergas with a thick mold.
Any more than 5 seconds of CO2 gas on a SS core/mold, and you have ruined it.

A few years ago, I found out that SS can be set using a catalyst, and so I actually purchased some catalyst, but have not tried it yet.
The catalyst will cause all the sodium silicate to harden at the same time, throughout the mold or core, just as the resin binder does.
Sometimes with CO2, it can be difficult to inject into deep molds without missing a part of the mold.

I visited a local art-iron foundry in town, and noticed that they had both resin-bound sand and sodium silicate bound sand materials.
I asked them if SS could be used with iron, and they said yes, that the surface finish with SS sand was almost as good as with resin bound sand.

The thing to remember is that the bound sand only has to last long enough for the exterior of the metal to solidify.
Resin burns too, and it creates a charred interior of the mold, which I think actually helps with surface finish.
Since the binder is a small percentage of the sand, then the main point of contact between the molten iron and the mold is the face of the grains of sand, not the binder.

If a mold is made of some material that is too hard, supposedly it can cause a casting failure; not sure of the exact mechanics of that.

A buddy of mine makes his own sodium silicate, but I am not sure how.

.

 

[Richard Carlstedt]

Dazz- Yours is a big order and I admire you for the undertaking of it.
Be sure to record your work. Years ago I built a cupola and poured iron for my model steam engines.
There is much to learn about iron and fluidity has it's tricks . I was guided by a real iron metallurgist at the time who showed me some ropes
Are you going to pour it yourself or have a Commercial or Art Foundry do it for you ?
( pardon me for not reading all the threads to date)
My first comment is you should probably look at Silicon iron and definitely use a Vanadium additive. This will improve fluidity and machining immensely ( Vanadium) . Years ago this additive which is put in just before the pour was called "Hot Shot" as it creates a exothermic reaction ( sparks) and raises the temp of the pour which adds to being able to pour thin walls.

Remember 'draft" is only needed for withdrawing a pattern . I like the multipart sand molds shown in the Bugatti video and think you should consider using core molding sand (baked ) to assemble your mold.
The more complicated the part , the less "quick and dirty or easy" effort is required .
Dynamic changes in part thickness are a bugaboo ( Like going from a 1/2" thick to 1/8" wall )
Flow becomes an issue , as the iron cools going through the mold and if it hits thin part , will be too cold, so how the iron flows is important.
You probably know these factors already , but your thread made me remember issues i encountered
Good luck
Rich

If you use a foundry - be sure to spend time "early" with them to get the parameters you need !

 

[GreenTwin]

The guy who built the V8 in iron resorted to using scrap that had phosphorus in it, which is traditionally the material added to iron that is used for radiators and other intricate shapes that need extreme fluidity to fill completely.

The "books" say that for engine work phosphorus iron should not be used, due to problems with strength (if I recall correctly).
Obviously the V8 engine build proves that phosphorus iron will work on a model engine scale, and so the phosphorus scrap iron may be one option, as opposed to an exotic additive that may be difficult or impossible to source.

I am not going to resort to using phosphorus iron unless I get really desperate, but I think I can fill any molds I need to make without it, by using the correct pour temperature and ferrosilicon.

The base of my Frisco Standard engine is almost identical to the base of your engine, and I think I can get the base of the Frisco Standard to fill with multiple large runners in strategic places. (screencapture below).
The key to thin fills is long knife gates. You can see this in the work that people do casting thin plaques.

Ferrosilicon does help quite a bit with fluidity and with machineability of thin parts, and it is readily available from most foundry supply houses.
Ferrosilicon is not absolutely necessary for parts 3/4" thick and thicker, and I have cast parts without ferrosilicon, and they machine well as long as the part is perhaps 3/4" thick or thicker.

Having your burner tuned correctly is also critical to getting a very hot iron melt.
Getting iron melted and getting iron to pour temperature can be two entirely different things.
If the iron is slightly on the cold side, mostly likely it will not fill the mold completely.

I would also omit any filters when attempting to cast thin parts in iron.

It is easy to add superheat to aluminum, which is temperature above the pour temperature, and actually very easy to overheat aluminum.
Adding superheat to iron is not as easy with an oil burner, but I can be done with a properly tuned burner.
I have an optical iron-rated pyrometer, and I need to get it out and try using it again with my new furnace.
I would guess I am pouring in the 2500-2600 F range below.

You can tell when the iron is really hot because it begins to shoot out sparklers, which are tiny blobs of molten metal that look sort of like mini-fireworks.
I circled in green one of the sparklers from an iron pour where the metal was quite hot.
Also the reason you need to wear good leathers or similar protection during a pour; the sparklers fly in every direction.

 

[Richard Carlstedt]

GreenTwin, Very informative post !
I forgot about the knife gates .
Didn't know about the Phosphorus application .
Used "Silver" Coke ( The best) in the Coke Cupola and had to run my forced air supply through a 3 inch SS Tube first, that was buried in coals in a Weber grill to preheat the air for hotter pours. The Hot Shot kicked it up 300 degrees- added to the crucible at the moment of pouring .
Yes, Hot is good and Hotter is even better for good pours .

Tip from the metallurgist -- to read the machine-ability of the Cast Iron part even before touching it with a machine tool , do this
Make a Flat Wooden Wedge (pattern ) like the wedge in a old hammer handle
Make it 3 inches long and 2 inches wide and taper the long dimension from 3/8" thick , down to 1/16"
Put a "V" groove on each of the long sides. make sure the V's match in location.
When ready to pour your Iron, stick the wedge into some sand and remove it.
(you have made a miniature mold )
Now pour your regular mold and then pour the wedge mold
When it is cold, place it in a bench vise horizontally ( 3" wide x 3/8" deep) with the groove at top of jaw and hit it with a hammer, breaking it into two wedges 3 x 1" .. now "Read" the crustal structure at the break.---"White" iron is bad ( Hard) . This is a gauge of what to expect if you were to machine the part . My early pours had white halfway up the break ( 1 1/2") and later when I got to less than 1/2" white, I knew I had it . My first pours were too cold and the iron frooze to fast creating hard spots
The exact size of the wedge ( ~3") means nothing - the V allows the wedge to break at a predetermined spot ( weakest)

Rich

 

[GreenTwin]

The problem with the wedge test in a backyard setting is that unlike an industrial foundry, there is not a large furnace/ladle full of iron to sample from that will easily remain at pouring temperature.
I guess you could leave the crucible in the furnace during a wedge test.

But the wedge test seems counterproductive because in order to check the wedge in real-time, the wedge would have to be cooled quickly after it was poured, and cooling iron quickly is the best way to get chilled spots in it.

Any iron parts poured should remain in the mold overnight, and allowed to cool as slowly as possible, otherwise any thin parts will not be machinable.

I guess you could pour some wedge tests, and let them cool overnight, and then test them, but that wastes a lot of fuel, and wear and tear on the crucible.

Another test I have seen is the spiral test, and it measures how far along a spiral the iron will flow.
The spiral test could be used as sort of a chill and flow test.

With the correct amount of ferrosilicon, the chills can be eliminated (I am not having chill problems).
Too much ferrosilicon, and there will be excessive shrinkage and hot tears.
It just takes a tiny amount of ferrosilicon, but that tiny amount works wonders for eliminating chilled spots, and helps a lot with fluidity too.

Comparing a cupola iron pour and a crucible iron pour is not really comparing apples to apples, although they are similar.

There is about a 30 second window after pulling the crucible out of the furnace to pour iron while maintaining a good pour temperature.
The iron goes cold very quickly once the crucible is out of the furnace.
For thin parts, pouring as quickly as possible is essential.

Pat J
.

 

[GreenTwin]

Rich-

You got a sneak preview of the green twin at NAMES 2019.
It was not quite finished at the show, and I must say it did not seem to impress the crowd because only about three people looked at it.
Most of the work at NAMES is museum-grade, and so the humble little green twin did not stand a chance at getting any attention.

Before completion, at NAMES 2019:

After completion, published in Live Steam 2021:

And the man himself.
It is always humbling to be in the presence of one of the great masters of steam engine modeling.
If you have seen Rich's work, you will understand what I mean.

 

[GreenTwin]

2019 was my first visit to NAMES, and boy and I glad I made that trip because who knows if and when things will ever return to normal enough to have these shows again.

I was surprised to run into my casting buddy who does an aluminum pouring demonstration at the Soule Live Steam festival.
I said "What are you doing here?".
He said "What are you doing here?.
I said "Doing the modeling thing".

He is from Louisiana, and so had an even longer drive than myself (on left in photo).

 

[GreenTwin]

Here is John at the Soule Live Steam festival, with some of his jumbo coin castings.
He gets a superb finish on his coin castings, and typically like the one on the left in the second photo.