Parametric Filling System Model

I've been working in OpenSCAD (https://openscad.ord) - a code based CAD modeller - to produce a parametric filling system model. That is, you specify the parameters of the mold, metal, flask, and pour constraints, and the model will produce the "perfect" filling system design that can be exported and 3D printed.

The model design is based on the math from John Campbell's work, which I've distilled into an attached reference document.

The model handles two and three part flasks of any dimension. It includes the properties of the most common metals will calculate the volume and weight of the filling system, filling time, and minimum sections that can be filled for the available head pressure. You can even import an STL of your pattern to visualize the layout of the mold.

The model output consists of a filling basin, tapered sprue, runner, spin trap, gates, and feeder. When exported as a 3D printing file (STL or 3MF) it is broken up into parts to fit the specification of the 3D printer.

You will need to rename "Filling System.scad.txt" to just "Filling System.scad" due to file type restrictions of the forum.

 

Dean,

I'm trying to understand the spin trap function. It's at the end of the runner system. How does it help except for the initial shot of metal down the feed system? Wouldn't the "spinner need to be before the gate to continue to filter out gas bubbles and trash??

I have included a freeze trap in this runner system. The initial shot of metal with all the oxides/trash/debris winds up there, and freezes off. Then the gate starts to fill. Any bubbles wind up breaking the surface and going out the main riser/vent. This casting turned out quite nice. There was some slight porosity in the center thick section under the main riser. I'd like to eliminate that. In the new design that area isn't quite as thick. One thing I could change is to keep the runner from increasing in cross section until it gets to the gate. The pour basin/runner gate, is otherwise like you have shown.

I have Campbell's book on order, and haven't read it yet, so maybe I'm way out in left field here.



Monty

 

Yes, the spin trap is at the end of the runner to capture the initial shot of metal until the sprue and runner are full. The theory is that the turbulence of the initial filling of the system contains bubbles, loose sand, and oxides. The 'spin' part is to prevent a reflected wave from returning that trash back into the runner and gates. Once the system is initially full, the sprue and runner design is intended to minimize turbulence and keep the flow constrained so that no more bubbles are entrained. And if you are pouring in more trash then that's a problem to be solved elsewhere.

The runner should remain the same from the bottom of the sprue to the first gate. If you have more than one gate, then the runner should reduce in size accordingly to maintain flow rates and head pressures into the gates - though the preference is to divide the runner flow between the gates, rather than progressively draw from the runner at each gate.

 

Hi Dean, trying the SCAD file but cannot get it to compile. It looks wonderful, I have only ever gotten through a dozen lines of code in OpenSCAD and most of that was by borrowing bits and pieces I saw elsewhere...

OpenSCAD version 2015.03-2

ERROR: Parser error in line 151: syntax error
ERROR: Compilation failed!

151 basin_reservoir_depth = (average_fill_rate / molten_metal_density[metal] * 1000 * basin_reservoir_time) / (PI * (basin_diameter/2)^2);

Doesn't seem to like the 'carat' [^]

 

Up and running in OpenSCAD 2021 with no changes to file posted.



With one of my STL files, the sizes of the features change as variables are modified:


Impressive work Dean!

 

Well, there's ya problem. An OpenSCAD version from 2015 is going to be missing a few features. It's under continuous development and the latest stable version is 2021.01

 

I've made a couple of changes to make it easier to print. The sprue and spin trap vent now have a flat side so that they can be printer on their side, rather than needing to print a tall, very thin part.

Updated files are available from https://www.printables.com/model/235393-parametric-filling-system-model-for-metal-casting

 

Thanks Dean. I forgot to mention tried the previous rev from design through 3D printing. I poured with a little heavier sprue and runners and vents on the far end of the part, no fillers or risers. Mocked-up with hot glue below. Looking forward to trying the new version.

 

Looks good!

But you've got the spin trap upside down - the conical part should be at the top with the vent attached.

And it looks like the gates are upside down as well - the curves should be away from the sprue. The idea is that they discourage the initial flow down the runner from entering the gates until the runner is full.

 

LOL. It figures I'd have the half of it backwards...probably should have printed it out. The why's' help it make sense too, thanks for taking the time to explain.

 

I guess the two vertical square parts on the long in-gates are risers? My question is: Given the markedly less mass of the gates compared to the two "legs" of the casting, won't the segments between the risers and gates freeze long before the casting itself? So, how do those "risers" help? Or am I missing the whole point somehow? Perhaps calculations show that the sand is heated by flowing metal prior to the casting filling that the gates will still be communicating with the part. And, yet, even if they are those spindly risers should freeze moments after they fill. I don't get it.

And the pouring basin seems so large---why?

My overall impression is that this whole setup is long spindly and mighty complicated. Especially since the part was successfully filled using a much more compact and simpler setup. But, I would be interested in 30000-foot explanation if someone cares to provide that. What is the fill/freeze sequence envisioned that casues this proposed setup to work?

Denis

 

I'm with you Denis.
From the most basic setup that worked to a piece of complicated geometry.
I did alter a couple of my practices after the discussions about Campbell's work.
One was get rid of splash wells at the bottom of the sprue, the other was a heavily tapered runner.
Spin trap...maybe if the pattern was 10 inches deep, I might consider it. But only after a failure that I could attribute to squirting metal.
I subscribe to the KISS principle and it most often works.
Edit: I also added a pouring basin as per Olfoundryman.

 

Denis, some of the variables I entered were numbers I thought about to fit my flask, many were more along the lines of uneducated guesses.
So 'my' design is probably not optimal to either common practice or to Campbell's theories.
I wanted to acknowledge what Dean did by running the OpenSCAD session and 3D printing it.
Dean would be the best person for the summary, some of it may already be in the PDF presented earlier. I am just fiddling with the knobs and buttons.

 

Yes, they are risers and yes, I agree with you, they most likely won't work like that. In fact Campbell recommends against risers on the gates.

Proper placement and sizing of risers is a whole separate calculation that is dependent of the properties of the piece being cast.

The pouring basin is calculated based on the diameter of the area that you want to pour into and the reservoir capacity in seconds. The reservoir capacity is how long the basin can empty before the top of the sprue becomes exposed. The default is one second, with a full basin you can stop pouring for 1 second before it drains down to the sprue. Now the purpose isn't to allow you to stop pouring mid-pour, but to act as a buffer so that you can react to changing flow rates as you pour.

That is a fair observation.

One of the design parameters is casting fill time based on a ideal vertical rate of rise of the metal in the cast. For Al this is 60mm/s. Campbell's guidance is that this should be as slow as possible. From that and using the casting height and volume, it is possible to calculate a desired flow rate and size the runner and sprue accordingly.

One of the criticisms of Campbell's work when applied to small castings is that the sprue and runner are very long and spindly.

The example dragon casting shown in the pictures in my first post is 190mm in diameter, 20mm high, but only 130ml in volume - that's about half a cup of metal.

Now if you pour half a cup of metal out over 5-10 seconds and look at how thin the stream is - that is the size of the sprue and runner if they are to remain full during the pour. In the case of this example, it is 5mm x 5mm after a 90mm drop, which is about what you observe.

Like you said, the part was successfully filled using a much more compact and simpler setup, and that is the experience of many casters over many years. It is also their experience that the quality of their castings is often poor, if they are even concerned about that at all. (Though in reality, hobby casters have become accustom to poor castings and are accepting of them). If what you do works for you and you are happy with the results, keep doing it.

What this model does is it puts the maths of Campbell's theories into practice in a way that can be experimented with and tweaked by fiddling with the knobs for small castings.


Dean

 

Dean,

I went back and looked at your original post modeling the dragon and can see that the proposed feeding system seems more intuitively reasonable than the one for the sailboat part. (I also now understand that Tops was not suggesting that as an ideal system, but more just a matter of seeing if he could generate an output to approximate a system that would potentially be tuned for the part in question.)

I do not know how much trouble it would be to plug in numbers for the sailboat part and look at the basin/runner/gate design. If it is easy, it would be interesting to see it.

I like large pouring basins for use in my (mostly) iron foundry work for a couple reasons. I pour using a trolley with a 7 foot beam between the handles and the crucible. So, I need a large target to avoid slopping metal. And I do very much like having a good-sized buffer which makes it easy to keep the basin full throughout my usually about 15 second pours. One non-volume dependent refinement for me has been to use a basin with in-sloping side walls (tumblehome in nautical design terms) which really helps the initial unavoidable splash of metal entering the basin and spreading laterally to be curled back into the basin and not so likely to shoot up the side and onto the top of the mold. Once the basin is full, the metal itself dampens sloshing.

It sounds like the work you’ve done to make Cambell’s work more accessible to small-scale foundry workers is a bonus.

How much have you utilized this system in practice? How often is material like sand or slag actually identified in the spin traps?

Denis

 

Changing the variables for cope height, basin diameter, basin reserve capacity, mold fill time, and whether or not to enable feeders make for a different result:


I suppose in industry one would do a design-of-experiment (DOE) and figure out which variables, when changed, lead to more favorable results.

 

That looks so much more reasonable. In practice, I think I would be inclined to cut the vetical ends of the "ears" of the casting so that they sloped downward45 degrees or so just to avoid a waterfall of aluminum while the mold begins to fill. They would be very easy to saw off at a right angle later. And I am puzzled as to why the sprue is rectangular when ease of formation, surface area, and smoothness of flow would favor a round sprue. The runners are practical to cut as rectangles or semi rectangles. The "tower" on the spin trap is supposed to trap debris and flow it upward, I guess?

It does look like that would fill. But, it still probably will need a riser on each ear near the gate (my guess).

Thanks for posting the revision.

Denis

 

Schematic from the second pour, there was a small shrink defect like 1mm (.040"), hard to see in photos, at the marked spot. Image does not have the fillets added in my 'pattern rapping fail' thread.

 

This is where things are a bit counter intuitive. You actually want more surface area to increase the friction to slow down the flow of the metal. One of Campbell's design objectives is to keep the speed of the metal flow low. Also, if you have a square runner, then a square sprue decreases the turbulence at the sprue/runner junction.

The spin trap is sized to contain the entire filling system volume so that the filling system is full of 'fresh' metal before the casting starts to fill. The conical shape is intended to gradually increase the back pressure rather than cause a shockwave effect when the metal hits the top. The vent is the same cross section area as the runner to avoid any premature back pressure buildup - also tapered to provide a draft.

Dean