How to CNC machine a pattern for sand casting
Casting a steering wheel for a Go-kart, at NTNU University
One of the places where DeskProto is used in education is at the Norwegian University of Science and Technology NTNU in Trondheim. Their department of Product Design uses DeskProto and an Isel milling machine for Rapid Prototyping of the students' design projects.Some of their students venture into very complex projects, as shown in the example on this page. The project is done by student Alex Mitchell, and includes the complete manufacturing process using sand-casting.
![The original Go-Kart](../photos/ntnu-go-kart.jpg)
The Go-Kart with the original steering wheel, which needs to be improved.
Starting point for this project was the "Go-Kart" that a Mr. Knut Sørby has made for his children, shown in the picture above. It was felt that the rather crude steering wheel did not match the design of the cart, and an assignment was made to design and manufacture a better wheel.
The wheel needs to be designed so that it can be attached to an M8 threaded rod. Required total diameter is about 230 mm.
![Car from the (19) twenties](../photos/ntnu-orig-car.jpg)
![Steer of such car](../photos/ntnu-orig-steer.jpg)
The Kart was inspired by cars like this one, so the steering wheel should be as well.
The inspiration for the design of the cart came from automobiles that were made between 1900 and 1930. So it made sense to also base the design of the steering wheel on the wheels that were used in these cars.
![First concept design](../photos/ntnu-design-1.jpg)
First concept design.
A number of concept designs have been made and evaluated. One of the iterations for instance involved a change from straight spokes to curved spokes, needed to accommodate the shrink after casting (straight spokes may crack or tear when cooling down).
![The final design](../photos/ntnu-design-final.jpg)
![Schematic mould design](../photos/ntnu-mould-design.jpg)
The final design (to be produced) and how this wheel is positioned in the mould.
The design of the mould and of the running system also involved some iterations. The issues to be taken care of included:
- Filling at the required speed (too slow -> incomplete fill; too fast -> bad structure).
- Delivery of only liquid metal (trapping of any inclusions).
- Eliminate turbulence.
- Ease of removal.
![Side one of the pattern](../photos/ntnu-pattern-design-1.jpg)
![Side two of the pattern](../photos/ntnu-pattern-design-2.jpg)
The design of the patterns (side 1 and side 2), with parting surface and positioning pins.
The mould as drawn above will be made in sand, in two halves, and to create each halve a pattern has to be machined. Each of the two patterns shows one half of the geometry (positive shape).
Where no geometry is present the two mould halves should fit, so this "parting surface" needs to be an exact match at both patterns. As you can see between the spokes a 3D curved parting surface is used (for injection moulding DeskProto toolpaths are not sufficiently accurate for curved parting surface).
![DeskProto screenshot](../photos/ntnu-dpscreen.gif)
![The CNC milling machine at NTNU](../photos/ntnu-milling-machine.jpg)
Pattern and toolpaths in DeskProto, and the CNC milling machine that has been used.
Toolpaths have been calculated using DeskProto, and the two patterns have been machined in Renshape on the Isel GPV 4830 machine of the Design department.
![The machined pattern, side 1](../photos/ntnu-pattern-side-1.jpg)
![The machined pattern, side 2](../photos/ntnu-pattern-side-2.jpg)
The two patterns, made a bit larger after machining by adding extra material.
The two machined patterns that have been machined proved to be a bit too small to produce correct moulds, so some extra blocks of material have been attached.
You can clearly see three positioning pins on side one, and the three corresponding holes in side two. These are of course used to exactly align the two mould halves. Also part of the feeding system is visible. The rest of the feeding system, and also the riser in the center will be drilled into the sand mould later.
![Adding sand](../photos/ntnu-sand-filling-0.jpg)
![Compressing sand](../photos/ntnu-sand-filling-1.jpg)
![Placing the frame](../photos/ntnu-sand-filling-2.jpg)
Adding sand, compressing sand (side 1), placing the frame (side 2).
After placing a wooden frame around each pattern the moulds could be created. This is done by manually compressing the sand in the frame, using a special type of bonded sand. Sand casting like applied here is in fact one of the oldest manufacturing technologies.
![Assembling the mould](../photos/ntnu-mould-assembly.jpg)
![Casting the metal](../photos/ntnu-casting-process.jpg)
The two mould haves are combined, and next the wheel is cast.
After removing the pattern from each mould half, the two halves can be combined. The result is a large "block" of sand, with an (invisible) cavity inside that has the shape of the wheel.
Next the metal is melted in an oven and poured into the cavity. The mould halves are securely clamped together as otherwise the top half could start floating the heavy molten metal.
![After opening the mould](../photos/ntnu-casting-in-mould.jpg)
![The rough casting](../photos/ntnu-result-rough.jpg)
After opening the mould, and the resulting casting, without any finishing.
The next step is of course a long delay for completely cooling down both the wheel and the sand. After that the mould can be opened like shown above, or the casting can be removed by simply sifting the sand.
You can clearly see the feeding system and the riser, and (less clearly) some "Flash": a thin metal sheet where molten metal has escaped between the parting surfaces of the two mould halves. All this excess material has been removed using saw, file and grinder.
![The cast wheel after finishing](../photos/ntnu-result-finished.jpg)
The cast steering wheel, after finishing.
The resulting steering wheel is conform the specifications and can be used in the Go-kart. For future projects improvements are still possible in the surface quality: prevent defects caused by debris in the cavity, and use finer toolpaths (set a smaller Distance between the toolpaths in DeskProto) to create a smoother surface for the patterns.