Spot-welding to reduce delamination.

[leotilson@gmail.com 0]

3D printed parts are stronger in the X and Z direction than the Z direction. It has struck me that there may be a simple way to improve layer adhesion.

 

After printing, say, the fifth layer of a model, the nozzle should make another circuit just inside the inner skin, but, instead of printing a continuous length of plastic it should stop every 5 mm or so and extrude sufficient plastic that could run down the inside of the wall to a depth of 5 layers or so. In this way there would be strands of plastic running in the Z direction as well as X and Y. During this circuit it may be better to have the fan switched off.

 

At the sixth and subsequent layers this process should be repeated, but offset slightly at each layer.

 

A further improvement may be if the infill was arranged so that there were little channels for the plastic to flow down.  In the attached picture. for instance, there is a small channel which could guide the molten plastic.

 

 

[GregValiant 1,142]

I can understand the thought, but I have doubts about it being effective.

 

The extrusion starts cooling within micro-seconds of coming out of the nozzle.  It would be pretty hard once it got to the 2nd layer and I don't see any real bonding beyond that.  There isn't any squish like between adjacent layers either.

 

When mechanically welding plastics both the pieces are heated, the heater is rapidly retracted, and the two parts are forced together.  The bond is made when the near liquid high temp plastic on both pieces fuse and then cool together.  It's the same sort of thing in friction welding and spot welding steel.  In MIG and TIG welding, the arc melts both the wire and the workpiece.  In FDM the hot extrusion would immediately dribble down a (relatively) ice cold workpiece and there would be no force to keep it against the workpiece.  The "little channel" approach might have the best chance of working.

 

I'm sure no one was going to ask me, but I want to state for the record that I don't want any part of being on the software design team for that.

[Longtoke 25]

Interesting Idea, and I'm sure I've had some printing mishaps that essentially replicated this process with the resulting "model" being stronger than you would think, but the end result won't be consistently reproduceable. Once you start talking about letting the filament flow freely, it almost always does random things. 

 

What you are describing ( better Z layer adhesion ) is already available of sorts ( in a limited and different way ) with variable layer height using a different slicer. In Cura we have "adaptive layers" which allegedly gives a similar result, but being listed as an experimental feature, I can't say I've even looked at it. 

 

Another avenue to look at for increasing the strength of prints is post processing with annealing. My own tests with this have bought mixed results, the biggest issue being shrinkage of parts, in some cases as much as 5%. 

 

Food for thought either way. 

[nallath 1,118]

This is what I applied a patent for (or well, not quite, but it's close).

 

My idea was to actually move the nozzle down through a number of layers. This makes the layers less planar and re-applies heat as it moves down, improving the layer adhesion in z direction.