E.P. Allis 3000hp Quadruple Expansion Corliss Engine, 1893 in 1/12 scale

[Dave Otto]

Nice results!

Dave

[bananarchy]

Flywheel Spoke Machining - End Faces

After a few more pours, I've got fourteen (potentially) good parts and two setup parts ready to go on the mill. A couple will end up having more shrinkage porosity than I thought, so I'll have to cast a couple more, but I'm proceeding with this set for now, since it's going to be a bit before the next pour.

The mating ends of the raw castings are about 1.400" thick, and the objective is to finish the hub end at 1.000" and the rim end at .750", with the cast shaft of the spoke located symmetrically between the machined end faces. To achieve this, each spoke ended up going on the mill table 7 times and the surface grinder 4 times. There was probably a way to reduce that count slightly, but results matter more than efficiency in this case. Both sides of both ends were roughed down leaving .050" per side, and then measurements were taken on the surface plate to see how asymmetrical the shaft was to the rough faces. The shafts are rounded and tapered, so a consistent location was marked from each end, and the measurements in this spot were compared when the part was flipped over. Each part was marked with the necessary adjustments, and material was removed as needed from each end, with shims placed as needed under one end on the second cut to ensure everything is properly parallel. Some parts required no adjustment at all, and the worst was about .035" out. Not bad.

Following centralization, the ends are brought down to the final (mill) dimension, leaving 5 thou per side for grinding. Clamping these on the table was a little tricky - I've got the aluminum bars running across the spoke shafts, but any real force on these bends the shafts and distorts the geometry of the ends, so the main constraint is with toe clamps from the sides. Not my favorite, but nothing moved.

Both sides (and both ends) are then ground to 1.001" for a nice snug fit with the hub (the width of the mating groove ended up a grand or two above nominal). Four parts fit nicely on the 8x18" mag chuck.

[bananarchy]

Flywheel Spokes - End Faces Cont

Now with both ends at 1.0", only four steps left - mill, grind, mill, grind on the rim ends of the spokes to get them to .750". The first side repeats the previous milling setup, with 5 thou left for grinding (to a total thickness of .875"). For the second side, a spacer is required between the spoke end and the table, so an old parallel was sacrificed to the gods of precision and ground to .1255" (the extra half thou is there because the hub ends are actually 1.001"). After one last trip to the mill, the final side is ground in batches of 3. This is the first real job I've done with this grinder and I'm still learning a bit, but having the right wheel definitely makes a big difference (Norton 32A46-IVBE). I was able to hit the thickness +/-.0002" no problem, with most of the parts within .0001". Grinding really helps with getting everything nicely flat and parallel, since there's basically zero distortion due to workholding forces.

Next up, drilling a bunch of holes!

[CI]

The castings look very good.
A few thoughts about risers, runners, gates, sprues, etc.:
I can only speculate; I don't know exactly what would or would not work in your exact situation, so take all this as what if's.

1. I normally use a rectangular gate at the top of the runner, so the runner would be in the drag, and the gate in the cope.
The idea is that when you pour, the runner fills first, and sweeps slag, inclusions, loose sand, entrained air, etc. into the spin trap before the metal completely fills the runner.
Once the runner is full of hot clean metal, then the gate starts filling the mold cavity.
If the runner and gate are at the same level, some trash will get swept into the gate as the initial metal flows past the gate.

2.  You could use a sprue at one side where one of the spin traps is located, and let the runner feed a single spin trap on the opposite end of the runner.
This would allow you to use one spin trap only, and stil get the sweeping/cleaning action.

3. The abrupt 90 degree transition at the bottom of the sprue where it meets the runner will cause a splash back when the sprue is filling, entraining air, slag, etc.
The bottom of the sprue could have a smooth radius transition into the runner to maintain laminar flow.
The spin trap will take care of the initial turbulence and entrained air/sand/slag, but the turbulence during due to a 90 degree bend may break off some sand.

4. Risers are always a bit of an art.
If I were doing it, I would have a 2.5" diameter, 4" tall riser above each end of each spoke, with a neck at the base of the riser into the casting about 1" diameter.
Risers tend to push the crucible size up quite a bit.
Having large sections on either end of a smaller section would be prone to hot tears without risers.

I have seen some use a runner down the center of two long castings, with gates branching off to the mold cavities in two or more places, with the sprue in the center of the length of the runner.
Again, I would put the runner in the drag, and the gates in the cope at the top of the runner.

Your castings are so good that one can only speculate on possible ways that may or may not help to improve them.
Looking great !
Following along.
 

[bananarchy]

Flywheel Spoke Drilling - Initial Attempt

As with any castings, correctly positioning the machined geometry relative to the casting is a critical and potentially challenging step in the process. The first couple attempts didn't quite turn out right, as the geometric interactions between the components are kinda tricky, and getting everything to line up visually with the adjacent spoke in the hub was harder than I initially anticipated. A good bit of time was spent recently on other projects, but also just thinking about the location methodology and waiting until I came up with a method I could be confident would produce the results I wanted.

In addition to printing casting patterns, this is another part of this project where 3D printing comes in extremely handy. I'm able to model locating jigs which directly reference the cad model of the part itself, and iterate on the design with minimal expenditure of time and material. The basic idea of these jigs is to determine the location of the origin hole (the centermost hub bolt hole), which fully locates everything when combined with the part centerline. The initial attempt, done on a set of four test parts, relied on manually finding the centerline and eyeballing the longitudinal position to split the difference where the edges of the spoke pads interact with the hub and rim sections.

Once the centerline was scribed and the origin hole is center punched, the part is located in the mill using a wiggler and the holes are drilled and reamed. Next, the part is bolted to a fixture plate to machine the faces of the end for clearance. This worked nicely, since it only requires locating one corner of the fixture plate and machining to a known location.

It turns out that the eyeballing method wasn't reliable enough, and the alignment between adjacent parts wasn't fantastic.

[bananarchy]

Spoke Drilling - Final Execution

A secondary learning from the initial batch of parts is that there was a bit more variability than I like in the center to center distance of the bolt holes at the outer ends from one spoke to the next - this makes sense when you think about it, since any small variance in the hole locations at the hub end is magnified ~10 fold at the outer end. New plan - drill the outer holes on the rotary table as an assembly.

The locating fixture was updated to reference more geometry and take the eyeballing out of the equation. Pins were added to reference the pad edges to ensure a consistent cap to the edge of the hub, and a slider was added to reference the outer edges of the spoke shaft. Finding the middle of the spoke shaft was tricky due to the taper and casting variance, but this worked like a champ.

The full batch of parts was drilled, reamed and milled for clearance as above. I got a 5/8" carbide roughing mill recently and it will take some monstrous cuts in cast iron without the machine complaining at all - strong recommend. Then everything is test fit in the hub for the first time (just last night!). I've cut the stock for the studs to length, so it's the final material and size just without threads yet. Most went together with no effort or a light tap, one has alignment issues but given how tight the clearances are that's not bad at all. I'm extremely pleased with how the adjacent parts line up at the hub interfaces.

[bananarchy]

Spoke Drilling - Results

Here's the final assembly, and also a look at how it will interface with the rim segments on the test parts which had the rim holes drilled

[crueby]

Thats quite an assembly, the wheel parts look amazing!   

[Sanjay F]

Wow!! 

[bananarchy]

Flywheel Rim Segment Casting

After proving out the pattern and corebox strategy, a second version of the pattern was made that eliminates some of the extra material on the exterior face - there's still a flat on the part which will be convenient for machining, but there's less extra to have to turn off the OD, and the final surface will be closer to the cast surface of the part, minimizing the risk of encountering porosity on the OD face of the finished part. A sizable riser was also added. One initial test part was done which ended up with some gnarly inclusions, and then a batch of four proper parts was poured. Pretty happy with the results - #3 had something funky happen on one corner, which might or might not clean up, and there's a wee bit of shrinkage porosity in the riser neck on one or two parts. Dimensionally we're in good shape, and I'm very pleased with the surface finish.

[bananarchy]

Flywheel Rim - Orientation and Reference Face Grinding

The locating and machining sequence of these parts has been quite the puzzle - there aren't nice reference surfaces on the part itself, everything is angled, etc. The most important parameter, though, is that each segment lines up nicely with its neighbor at the end face where they bolt together. So that is what we need to locate the part relative to, but how? I designed and 3D printed a nest which cradles the part at four points of contact, which are the outermost corners of the faces I want to line up. The nest is in two parts which slide together, both to accommodate for part to part variation but also to allow the nest to make it past the overhang. When placed in the nest, the part is located in two degrees of freedom.

Now that we've got it located, the reference surface flat is machined in the surface grinder - this is a very time consuming process, but minimal cutting forces are necessary given the less than ideal workholding setup, and since the grinder is automatic it's not too bad. The nest includes a pocket in the center for some metal to help hold it on the mag chuck, which incidentally didn't work at all  so the part was blocked in thoroughly. This worked pretty well! The part was ground until the face cleaned up basically all the way across, rather than to a specific dimension.

To check the orientation of the setup part, lines were scribed based on the edges of the nest (which are close to the finished end faces post machining) and the height of those points was compared on the surface plate - ended up within a few thou! More than good enough.

 Next, that face can be clamped to a right angle plate and then one outer edge of the part skimmed to orient the part in the next DOF

[cnr6400]

 

[Chipswitheverything]

What you are showing and describing in this engine build, the casting, machining, the holding jigs and process involving very modern methods, is really remarkable, and at a pioneering end of model engineering. This will eventually be a most impressive and mighty model engine!  Dave

[pgp001]

That flywheel makes mine look like a baby at only a foot diameter.

[CI]

I was reading this very interesting thread, and was scratching my head, saying "You know, there are not many machine shops that have an iron foundry in them".
I like to keep up with who is doing what iron casting, and where.
Then I looked at the angle iron on top the flasks.
I only know of one iron foundry that does that regularly.
This thread has a K.S.W. feel to it for sure; it can be no other.

Great design and castings !

.