LPSE v4 failure during assembly, and an adventure in O-rings.

V4 of my 3D Printed Liquid piston Stirling engine is another failure.  This time for an unexpected reason.

After several failed attempts, I'm not able to print, retrieve, and install the new narrower TPU gasket.  The TPU sticks very strongly to the print bed, and the first time I tried to pull it off it ripped apart between layers.  In another print I was able to get the gasket off without tearing, but there were some surface deformities and stringing that, even after cleanup, prevented installing the gasket in the face slot.  Since I have an MMU, I feel reasonably confident I could print-in-place the TPU component, but that would make this project inaccessible to people without that capability.

A logical solution for this would be to make the gasket wider.  That should create an additional cross-section area with greater layer adhesion, and I could fiddle with the print settings to print it continuously to eliminate stringing.

At the risk of going down a rabbit hole, I’m going to try something different.  V5 will attempt to use the TPU filament as an o-ring and eliminate the printed gasket entirely.  I've never used O-rings before, so it's time for a side project.

My aging Machinery's Handbook (24e) came up empty on O-rings.  Turning to the internet, Parker has a 292 PAGE design handbook here: (free) PH_ORingCover.indd (parker.com)

Reading that was informative.  My earlier intuition was correct about how sealing works.  The pressure of the ring or gasket against the faces needs to exceed the pressure of the liquid trying to escape.  An O-ring does this by elastic deformation between the upper and lower faces.  Running through the design guide the critical parameters are that this needs to be a liquid tight static face seal using nominal 1.75mm TPU of 95A shore hardness as the o-ring material.  That is very close to the nominal size of 1/16 freedom units, so I printed a test sample piece (4a1) with grooves at my wild guess dimensions of 80% depth/130% width, plus the minimum, maximum, and median dimensions for a liquid tight connection suggested here: PH_5700_Sec4.indd (parker.com)

This was informative, and I'm glad I didn't just blindly drop the grooves into the main 2 hour print.  Following this design, the O-ring seals against the top and bottom face, and the width of the channel is significantly oversized to allow for high tolerance in O-ring fitment.  This is problematic with my 3D printed components because the print head leaves relatively high surface roughness on the bottom of the O-ring groove where the printed channel-bottom-top-layers are at a 45 degree angle to the groove.  I might be able to work around that by creating a radius on the bottom of the O-ring groove, ironing those layers, or (least desirable) sanding them smooth.  I closed the O-ring book and decided to take a different track instead.  Let’s try using the layer lines on the side of the groove as the sealing surface instead.  They are smooth (individually) and should work as sealing surfaces with my relatively hard O-ring material.

To nail down some dimensions I printed another test piece (4a2) to find a groove width that allows reasonably easy installation of the filament with a snug fit.  I found a nominal width of 1.575mm is a 10% horizontal squeeze on the filament O-ring.  The handbook says the total squeeze on a face seal is 19% to 32%, but again their O-rings are much softer than my TPU.


This looks like it will work so now I need to figure out the ring groove depth.  Test piece 4b2 was designed and printed with depths of 1mm to 1.4mm.

Here’s where I hit another 3-D printing specific feature.  I’m printing the model with 0.3mm layers, so the grooves printed with 1.1 to 1.3mm depths are all physically printed at the same height.  I could fix this by enabling adaptive layer heights or reducing the layer heights for the entire model, at the cost of longer print times and additional complexity.  This is, by far, the most fidgety and printer specific parameter of this design (so far!) and I’ve already strayed away from a proper O-ring design per the book.  I’m moving forward with a v5 body print and 1.2mm depth and 1.575 mm width O-ring groove.

P.s. I just shipped the job to the printer and realized I’ve exchanged the fiddly bit of printing a narrow high aspect ratio TPU gasket with the fiddly bit of gluing together the O-ring properly to prevent leaks around the cover screws around the island. $1 says this thing pees all over my workbench when I turn the heater on. ☹  I may have to take another go at this gasket business. :/