A strategy that I have undertaken is to make something similar to a family mold, but with a selector mechanism that directs the plastic to separate cavities. This way the tool cost is similar to a family mold, but without the balancing problem. I don’t know what the technical term for this approach is, but surely I am not the first person to have thought of it.
I agree. You will have no flow balancing problems. However, the center of area of the projected area will be offset from the center of the machine. This could cause flash and mould maintenance issues as well as higher clamp force (I don't really know how significant) especially if the offset distance is large. Make sure you have enough clamp force in the selected machine in case you need it. Also, if possible make the flow direction mechanism T shaped so that you can have the option of running one or both cavities at the same time.
I've been interested in aluminum molds, but my fear is that the mold is too easy to damage in production; the material is just too soft. Also, we have no way to grind it, which complicated tool making.
Makes sense to me. I would make some experience with non aesthetic, low production, easy flow material parts. If there are small core outs, they could easily bend. Keep the area clear of cooling lines in case you need to insert them in the future or just insert them from the beginning. I have run up to about 10,000 parts in aluminum tools but unfortunately never ran them until they fail or deteriorate. A mould maker told me once that they could guarantee 100,000 shots but I have never confirmed this.
.... In particular, is it better to machine the cavity into the mold plates, or cut out an insert which will hold the cavity?
Two reasons I can think of to insert the cavity in a pocket in the mould base: one, when the steel will be heat treated as heat treating the large cavity or core plate will deform them and two, for parts that will undergo important design changes in the future so that the insert can be easily modified or just replaced. Cutting the cavities right in the mould base is just faster and easier. I have also used them to reduce the cost of the prototype tool. The production mould base is built with the prototype inserts and later on, the production inserts mounted on the same mould base.
Also, what about three plate molds? It gives more flexibility in positioning the gate,
Other than the gate positioning benefit (is that an issue in your case?), three mould plates are used for automatic ejection and separation of the runner and gate system from the part. This is a benefit for high volumes production. Not your case.
...but as I see it requires at least two cavities to keep the mold balanced, ...
On the contrary. One of the applications of a three mould plate is to allow ONE cavity tools to have the center of area of the part in line with the axis of the clamp unit while being edge gated.
in addition to the expanse of extended screws and roller pullers.
Yes they are more expensive. They also could require a machine with a longer daylight as the shot height is longer and extra spaced needed for runner extraction.
What about mold insert units, as opposed to complete mold bases? It appears that the blank cost of an insert is comparable to the cost of a small mold base, so what is the advantage? Is it just about a quicker change over in the press?
I think that quick change mould bases have more application in high production high change over environments and where part sizes are similar. Because of your low volumes each production run will most probably produce parts for several months. If you have parts that are similar and can use the same ejection system (or easily adapted) I would make inserts and use the same mould base (another use of using cavity insert blocks in the mould base).