36x60x54 Twin Tandem Mill Engine

[Jasonb]

It is possible to add draft at the design stage. You will also have to add machining allowances to all surfaces that need it, again do it at the design stage. Then your 3D prints will have teh draft and just need the surface ssmoothing before they are ready to use as patterns.

My usual way is to include the draft as I go and then make a copy of that part which can then have machining and shrinkage allowances added to become the pattern.

You will also need to use cores for that bed pattern unless those cross head guides are separate., the core boxes that they are formed in can be derived from your 3D model

 

[GreenTwin]

I would recommend not making a copy of a 3D model, but rather create the model like you have done, which is to some extent the "as-machined" state of the engine, and then add features, such as machining allowances, and perhaps core prints.
The features can easily be suppressed in the model, to get it back to an "as-machined" state, which is what is used to derive 2D drawings.

Turn on the features if you are going to print a 3D printed pattern.

One thing missing from your model that will cause a lot of sand mold problems is fillets in the corners, and at the plane intersections.
I generally make the fillets a part of the "as-machined" 3D model.
I generally don't add fillets until the 3D model is completed and checked, because fillets will often malfunction if changes are made to the base shapes of the model.
From a casting strength standpoint, you don't want any sharp corners/intersections, since this is a stress concentration point.

You can get away with some lack of draft angle if bound sand is used.

As JasonB mentions, you can sometimes use the pattern itself as a corebox, if you add the coreprint parts to the pattern.
Coreprints would also be a feature in the model that would be suppressed as desired.
The coreprints are projections into the main mold which support the cores.

Some of the more exotic methods could be used to cast the parts, such as lost PLA, but with bound sand, you would not need an investment process (in my opinion).

Some situations may require retracts, or loose pattern pieces, such as the crosshead guides.

And as I mentioned, I create the 3D model at 1:1, and then scale the entire 3D model down in the slicer program (not in the 3D program). The shrinkage factor is added to the slicer scale, for example, if you 3D print your model at 1/2 scale, then a 12" long piece is scaled to 6" in the slicer, plus a scale factor of about 1.015 for shrinkage.
So the 3D printed part is about 6.09" long, and the casting should end up about 6" long.

The notes from the person who made the Merlin castings indicated that you many not always get the exact shrinkage in every casting, and he mentioned some alignment issues.
Larger castings are more tolerant of dimensional deviations from casting to casting, since you have more room for error.

I would recommend not adding draft to any surface that does not need draft; ie: don't fix things that are not broken.

The foundry and/or you need to determine pull directions, parting lines, etc.
The draft angle generally goes up to the parting line, and is not necessarily symmetrical in both directions.
Find the parting line before you do draft angles, if you choose to include draft angle in the model.

My first 3D printed patterns did not have draft angle, and I manually added machining allowance and draft angle.
I have since learned to incorporate draft angle into the 3D printed patterns, and while it is a bit more work, it is easier for me than manually adding draft to a 3D printed pattern later.
And the bonus of having draft angle incorporated into the 3D model is if you need to re-print a pattern for some reason (modifications, etc.), you don't have to repeat the manual draft angle work using fillers.

That the Y-shaped part of your frame is very similar to the Speedy Twin frame, and I have photos of the corebox and pattern for that if you want to see how they did it at the Soule factory.

Another reason to create the 3D model at 1:1 is that you can reproduce the original drawings for the engine.
This is the approach I am using for the Speedy Twin 3D model, and this will allow me to give the Soule museum an accurate full sized set of Speedy Twin drawings, which they don't currently have.

.
Edit 01:
Post #9 in this thread is an example of the first 3D printed pattern I made, which I then filled manually after printing.
These days I add the filling into the 3D model.

https://www.homemodelenginemachinis...engine-drawings-and-stp-files-by-pat-j.34335/

Some corebox and pattern shapes that resemble the frame in this thread can be found here:

https://www.homemodelenginemachinis...reprints-parting-lines-etc.36378/#post-416058

.

 

[Kawka777]

Hi All,
I have the engine pretty well laid out. It took a while to figure out the actual CL distance from each side, I did not have anything to go off of besides the crank, and that left a lot of questions.

I did make the bottom bearing “saddles” and the bottom main bearings. I change the journal to be 1”. The bottom bearing saddle will be held in place via a simple dowel pin and the main bearings will be held from spinning using a smaller dowel pin.

This was my first time looking at babbitt bearings that were designed to have multiple load zones due to the reciprocating motion. They actually made the sides of the bearing adjustable with a tapered key. Pretty incredible engineering. I still deal a lot of Babbitt bearings for large gear boxes which utilize a two piece bearing with a shimmable housing to achieve the right axial clearance.

My next step will be to make and install the crank. I need to make sure my engine CL’s are correct and I need to determine what the thrust surface thickness has to be on the main bearings to achieve the right crank position. Then I will go on to make the bearing caps and the top half of the main bearings.

Mike

 

[Kawka777]

I would recommend not making a copy of a 3D model, but rather create the model like you have done, which is to some extent the "as-machined" state of the engine, and then add features, such as machining allowances, and perhaps core prints.
The features can easily be suppressed in the model, to get it back to an "as-machined" state, which is what is used to derive 2D drawings.

Turn on the features if you are going to print a 3D printed pattern.

One thing missing from your model that will cause a lot of sand mold problems is fillets in the corners, and at the plane intersections.
I generally make the fillets a part of the "as-machined" 3D model.
I generally don't add fillets until the 3D model is completed and checked, because fillets will often malfunction if changes are made to the base shapes of the model.
From a casting strength standpoint, you don't want any sharp corners/intersections, since this is a stress concentration point.

You can get away with some lack of draft angle if bound sand is used.

As JasonB mentions, you can sometimes use the pattern itself as a corebox, if you add the coreprint parts to the pattern.
Coreprints would also be a feature in the model that would be suppressed as desired.
The coreprints are projections into the main mold which support the cores.

Some of the more exotic methods could be used to cast the parts, such as lost PLA, but with bound sand, you would not need an investment process (in my opinion).

Some situations may require retracts, or loose pattern pieces, such as the crosshead guides.

And as I mentioned, I create the 3D model at 1:1, and then scale the entire 3D model down in the slicer program (not in the 3D program). The shrinkage factor is added to the slicer scale, for example, if you 3D print your model at 1/2 scale, then a 12" long piece is scaled to 6" in the slicer, plus a scale factor of about 1.015 for shrinkage.
So the 3D printed part is about 6.09" long, and the casting should end up about 6" long.

The notes from the person who made the Merlin castings indicated that you many not always get the exact shrinkage in every casting, and he mentioned some alignment issues.
Larger castings are more tolerant of dimensional deviations from casting to casting, since you have more room for error.

I would recommend not adding draft to any surface that does not need draft; ie: don't fix things that are not broken.

The foundry and/or you need to determine pull directions, parting lines, etc.
The draft angle generally goes up to the parting line, and is not necessarily symmetrical in both directions.
Find the parting line before you do draft angles, if you choose to include draft angle in the model.

My first 3D printed patterns did not have draft angle, and I manually added machining allowance and draft angle.
I have since learned to incorporate draft angle into the 3D printed patterns, and while it is a bit more work, it is easier for me than manually adding draft to a 3D printed pattern later.
And the bonus of having draft angle incorporated into the 3D model is if you need to re-print a pattern for some reason (modifications, etc.), you don't have to repeat the manual draft angle work using fillers.

That the Y-shaped part of your frame is very similar to the Speedy Twin frame, and I have photos of the corebox and pattern for that if you want to see how they did it at the Soule factory.

Another reason to create the 3D model at 1:1 is that you can reproduce the original drawings for the engine.
This is the approach I am using for the Speedy Twin 3D model, and this will allow me to give the Soule museum an accurate full sized set of Speedy Twin drawings, which they don't currently have.

.
Edit 01:
Post #9 in this thread is an example of the first 3D printed pattern I made, which I then filled manually after printing.
These days I add the filling into the 3D model.

https://www.homemodelenginemachinis...engine-drawings-and-stp-files-by-pat-j.34335/

Some corebox and pattern shapes that resemble the frame in this thread can be found here:

https://www.homemodelenginemachinis...reprints-parting-lines-etc.36378/#post-416058

.

Thanks for those links to other threads. They are very helpful. I have been doing a lot of research lately on casting, but I am just getting ahead of myself right now. I need to get this engine in something we understand before I can do anything LOL.

 

[GreenTwin]

Looks great !
.

 

[Jasonb]

It's coming together well. Without looking back did you decide on the final scale?

I'm still inclined to say just a 3 casting engine (frames and flywheel) as getting the detail into the other parts will not be easy if cast. When you start to scale things down by such a large amount a different approach is needed than if you were making 1/2 or 3rd scale models particularly if you are putting the casting work out to a foundry where it may not be economical to spend the time that a backyard caster might. With the number of loose pieces (retracts) and cores needed on some of those items any slight misposition and you are going to start running into problems.

Also is it my eyes or are some of those stud holes very close to the edges, seems mostly on the LP valve "casting"

 

[Kawka777]

It's coming together well. Without looking back did you decide on the final scale?

I'm still inclined to say just a 3 casting engine (frames and flywheel) as getting the detail into the other parts will not be easy if cast. When you start to scale things down by such a large amount a different approach is needed than if you were making 1/2 or 3rd scale models particularly if you are putting the casting work out to a foundry where it may not be economical to spend the time that a backyard caster might. With the number of loose pieces (retracts) and cores needed on some of those items any slight misposition and you are going to start running into problems.

Also is it my eyes or are some of those stud holes very close to the edges, seems mostly on the LP valve "casting"

Hi Jason B,
Yeah all these stud holes ended up needing to be #0 on this. Do you have a standard that you all work off of in terms of how much material to leave between a stud and edge? I have head 1x diameter of stud. For larger bolts, 1.5x diameter material from a edge is standard.

Just to get through some drawings, I kept the center of a #0 stud a min of 0.083". That is only 0.052" of material which is like nothing lol. I made these sketches adaptive so i can easily change all that.

Mike

 

[Kawka777]

Hi All,
I decided to take some time to get some standard colors in the model and beatify it a bit.

It took a while, but I found, for sure, the axial CL distance between the two bores is 164". I had a ton of issues getting that right in the model, and found I missed a 1" shoulder on the crank I did not account for, which threw off everything. Now that everything is lined up, I went on to fit the rear cross slide.

The original cross slide is babbitted on all 4 sides. The bed plate had liners that bolted into the bottom where the cross slide rides and there are two gibs that hold the top down. The sides of the casting were machined to take up the sideways movement of the cross slide, but I could not find any reference to babbitt.

For the sake of this model, I am going to make the cross slide from bronze and allow it to ride directly on the machined casting surface (bottom). I also tossed around the idea of having a simple bronze wear liner on the bottom of the casting that would bolt to it. The gibs that will attach to the top will be notched to take up sideways loading too.

In doing all this, I was able to greatly simplify the LH rear bed. I added 1/4" radi in the corners of the cross slide area as I am expecting to mill the entire pocket with a 1/2" end mill. The holes for the gibs still need added. I did not touch the RH rear bed yet so you all can see how I changed it side by side.

I am a little hesitant to only capture the sideways movement of the cross slide on the top gibs, so if you all have any thoughts on that, please let me know. I cannot imagine there would be enough sideways force to cause any issues with this design, but you all would know a lot better! I added a sketch to what I am talking about.

MIKE

 

[Jasonb]

Ctr of stud wants to be something like 1.25 to 1.5 D in from the edge. But make sure any nuts don't hang over the edge, I assume you will be using small hex not standard.

A small edge thickness is more risky with iron than with a fabricated part as steel or brass is less likely to fracture at those sizes

Could only find sizes for steam chest covers but cylinder covers and flanges would be similar.

You could make the bottom liner a bit thicker and then mill a slot in it to give some sideways location. If the total width of teh slot is 1/2" then mill down the middle with say a 3/8" and then take a bit off each side to bring it to 1/2" which should be more accurate than a single pass which can pull the cutter sideways.

 

[GreenTwin]

If you will be running the engine with no load, slowly, as a model, then you can get away with things that you cannot get away with if you operate at a faster speed, with some amount of load.

Generally the crosshead is constrained one way or another in all four directions, ie: up-down-left-right, even though the bulk of the forces during operation may be either up or down.
.

 

[Jasonb]

just noticed you show bronze on bronze, it would be better to have the slippers in bronze and a steel crosshead or run steel on iron.

 

[GreenTwin]

You can run gray iron on gray iron also.
.

 

[Kawka777]

Jasonb, GreenTwin,
It makes sense what you said about the materials. I modified this to be a steel cross slide, riding on grey cast iron bottom bed plate and the top gibs are bronze but could be cast iron. I played around with the cross slide dimensions a bit and got something that may work. Here is a cross section view below.

I believe all those features in the bed casting would be easily machinable with an end mill. Added machine time to do it like this, but it makes for a nicer design than what I had before in my opinion.

 

[Kawka777]

Here is an exciting accomplishment on this project. The LP and HP pistons and rods are in and appear to move as intended. I had to play with the rod lengths to get everything right and even spacing at TDC and BDC.

I made the rods in two pieces and have them connecting at the intermediate cross head. The rear and intermediate cross heads have 1/8" taper pins to lock the rods in. That combined with retaining compound should be fine on these I think. The piston will be held in place via a 7/16" UNEF nut on the LP and a 1/4" UNEF nut on the HP.

I have the pistons and insides of the cap drawn to the print but are crazy complicated due to the form fit the pistons have in the heads. Eventually I will simplify them. However, everything fits up the way it should now. The cylinder caps need modified to suite the standard shaft rod sizes and whatever stuffing boxes that are needed for this size model.

I will be spending a while refining what I have, but soon will be valves, which I will definitely need your help with. The process is slow, I am getting digital copies of these old prints so those involved can see these drawings a bit better.

 

[Kawka777]

Hi all,
Here is an rough idea of how I think the valve system works for one side of this engine. I am not sure if the reversing cylinder is in the right place or anything.

I found there is a shaft called the "layshaft" is located 5'8" from the crank, located around 2ft above engine horizontal CL. That 5'8" is the PD of the large bull gear set. So the layshaft has a 5'8" bull gear in the middle and the 2 eccentrics on either side, one forward and one reverse. This is driven directly off the crank shaft as a 1:1 gear ratio. The confusing part of all this was rear bed drawings do not show bolt holes for attached castings, rather just a surface quality specification on the location of the fixtures. I imagine they would locate the holes in place when building the engine.

There is another drawing showing the HP and LP idler shafts and a casting that holds them, which is located just behind LP cylinders. The top shaft is labeled as the LP rock shaft and the bottom shaft has a linkage on engine CL, which must be for direct attachment to the HP valve.

Luckily from there, I used the attached valve motion diagram to rough out the idea.

The only unfortunate part is i can not find a drawing of the curved Stephenson link that the FWD and REV eccentrics attach to. That should not matter too much though if that is the only piece we are missing.

 

[GreenTwin]

Stephenson's link configurations varied; ie: the link could be curved convex (with respect to the crank), concave, or straight, and the rods can be crossed, nor not crossed.

There are also launch-style links, and I think what are called locomotive links.

The link suspension point is critical.

I am not sure if this is considered a "low-speed" engine (perhaps under 100 rpm), or a "high-speed" engine (up to perhaps 300 rpm).

It appears that there are only two eccentrics per side on this engine, but I have seen three eccentrics per side on the engine Jason linked, with the 3rd eccentric being perhaps a cutoff mechanism (riding cutoff ?).

Did we decide whether the valves are inside or outside admission?
This will affect the valve gear.
.

 

[GreenTwin]

Steam engines went through a long period of evolution, and one has to try to zero in on the particular time period for the engine in question, to find information that may apply.
I am assuming from looking at photos of similar engines, that this engine has a standard/typical Stephenson's link configuration, such as seen in this photo.

.

 

[GreenTwin]

Here are some pages from old steam design books.
Perhaps some of this can be applied to our tandem compound.

.

 

[GreenTwin]

The correct suspension point of the link is at the center, slightly offset towards the eccentrics a very small amount.
I have seen a white paper about how to calculate the exact point for the suspension pin so as to equalize valve travel in both forward and reverse gear positions, about like where it is shown in this illustration.

.

 

[GreenTwin]

Locomotive links have the eccentric rods attached outboard on the ends of the link, and launch links have the rods attached inboard of the ends of the link.

This would be a launch-style link.
The link should not be suspended from the end, but rather in the center, if you want equal valve travel in forward and reverse.
.