Spruing, plain and simple. Like most here, I've been mystified by
what is seemingly an impossible situation - normal, proper
investing simply cannot crush the model. I call it "internal
spruing" - the piece itself must have a way for metal to get
everywhere after it comes from the sprue rod. It takes a big sprue
to feed several small sprues - meaning internally.
A few years ago, a friend, now departed, helped me with a similar
problem. I was trying to cast some small pierce work pieces, about
0.70mm thick. One of the pieces is a lunette (half-circle)
approximately 30mm long by 13mm wide, pierced with 28 small holes.
The other is an outline only of a Celtic cross, with the outline
varying from 0.80mm to 1.2mm in width. The “wire” making up these
pieces is then roughly similar to filigree.
It took several tries, even with his expertise, before consistent
castings were made.
Each piece was connected with a wax wire about 1mm in diameter to
the end of a central sprue of considerable size - about 10mm to12mm.
This central sprue extended into the flask about 20mm in addition to
the rather large button. The models were placed in a single layer, as
if they were standing on end by their sprue. Great care should be
taken to ensure a smooth entry into each model as well as into the
central sprue. The lunettes had a sprue from each corner which joined
into a “Y” shape before being connected to the central sprue.
A section of wax mesh (such as used on the inside of a flask next to
the flask wall for venting purposes) was suspended 2mm to 3mm above
the wax models with vents running to the surface, i.e. the inside of
the rubber flask base. This mesh was not connected to the models or
the central sprue in any way. The theory is that air and other
gasses permeate the investment when the hot metal is forced in. These
gasses then collect in the cavities left by the mesh where they may
be vented to the outside, relieving pressure in the model cavity,
allowing metal to flow better.
This worked for me. I did several batches of each of 25-30 models
being cast. Only one or two castings were lost in each batch,
compared to over half in early attempts. The spruing process,
however, was like building a house of cards!
This may not work for you, but perhaps you might like to try this