After the somewhat embarrassing failure of our first Black and Tan siphon, a second attempt looked much more promising.  But would it actually work?

Background:

As we have previously mentioned, it should be possible to construct a Black and Tan by injecting a dense beer underneath a less dense beer.  Unfortunately, the previous design, when printed, contained holes that needed to be patched.  Once those holes were patched with paraffin wax, we realized that mistakes were made while slicing the model for printing which left support material mostly blocking the flow of fluid, preventing the siphon from working as intended.

Design:

As before, the basic idea is that there is a largish reservoir that sits atop a pint glass with a specially designed pipe that leads from the reservoir to the bottom of the pint glass.  When enough fluid is poured into the reservoir, a self starting siphon starts to empty the reservoir through a siphon pipe that leads to the bottom of the pint glass.  The siphon then runs until the fluid level reaches the intake of the siphon pipe.

Cross section showing the Beer Siphon Mk II sitting atop a pint glass.

The 3D model for the Beer Siphon Mk II is available on Thingiverse.

In the updated design, the bottom of the reservoir was thickened so it would print as five layers.  The edge between the bottom and wall of the reservoir was also rounded so layers would overlap slightly near the edge.  The pipe was also thickened to prevent leaks.

To lower the inlet of the siphon pipe as much as possible, the inlet was moved to the floor of the reservoir and the pipe was looped through the floor of the reservoir and back up to the new opening.

For stability, the three round feet were replaced with three long flat ridges along radii on the bottom of the reservoir.  This new design prints more smoothly as long linear strokes of the print head.

To allow for flexibility in testing, additional venting holes were added that could be covered with wax as needed to see which hole positions work best.

Finally, instead of boring straight pipe segments connected by circular bits, the new pipe is based on a golden spiral, which in addition to being more aesthetically pleasing should improve fluid flow by making direction changes smoother.

Testing:

Leakage and Siphoning:
The first thing that was tested was the reservoir’s ability to hold fluid (water) and the ability to start a siphon that would fully empty the reservoir.  For this, the siphon was placed over a pint glass and 8 fl. oz. was poured into the reservoir.  This revealed minor leaking from the reservoir, and was an insufficient volume to start a siphon.  By adding a few more ounces of water a siphon did start and mostly emptied the reservoir.  After a bit of testing, it was determined that by adding a single Jumbo Chilling Stone (2 fl. oz. in volume) to the reservoir, a siphon could be started with only 7 fl. oz.

Restarting The Siphon:
A bit of patching was done with paraffin wax to fix the leaking, and to cover the venting holes, which in some cases seemed to let air in and stop the siphoning early.  Once the reservoir was patched and cooled, a multi-stage test was attempted.  In this test 7 fl. oz. was poured into the reservoir and allowed to fully empty into the pint glass.  Then an additional 7 fl. oz. was poured into the reservoir.  At this point a siphon failed to start.  Starting a second siphon failed even after poking a small hole in the uppermost venting hole with a SIM ejector tool.

Layering Via Continuous Siphoning:
Having failed to start a second siphoning, one final attempt was made, using beer this time.  In this test Guinness was first sent through the siphon into the pint glass.  However shortly before the Guinness was done siphoning, Bass was slowly poured into the reservoir.  This prevented the siphon from finishing until the Bass had also been siphoned.  As shown below, this did result in a layer of Bass below the Guinness.  Unfortunately the Bass failed to fully drain into the glass due to air leaking into the pipe through the venting hole, breaking the siphon.

Partial success!  Guinness (upper) and Bass (lower) layers formed via siphoning. The second beer (Bass) was poured into the reservoir just before the first beer (Guinness) finished, so a fair amount of mixing occurred.  Some Bass also remained in the reservoir.

What went wrong:

1. Minor leaking was still an issue, but trivial when compared to the Mk I.
2. The reservoir was sized incorrectly, so more than 8 fl. oz. was needed to start a siphon.  However, Inserting a small metal cube into the reservoir allowed a siphon to start with only 7 fl. oz. of fluid.
   

   A small metal cube (normally used to chill whiskey) was used to reduce the effective volume of the reservoir.

3. Not using a support roof when printing resulted in a very rough surface on the interior of the reservoir.  As a result, some fluid ended up between layers in the floor of the reservoir.
   

   Reservoir interior after testing siphon using  water with blue food coloring. The rough surface caused some fluid to infiltrate the layers in the floor of the reservoir instead of draining properly.

4. Restarting the siphon did not work as intended, possibly due to the venting holes being too small.
5. In some tests the siphoning would stop, possibly due to air leaking in through the venting holes.

What went right:

1. Slicing improvements allowed the pipe to print without supports blocking flow.
2. The thicker siphon pipe wall did not require patching.
3. The radial foot design was much more stable.
4. Looping the drain pipe the floor of the reservoir worked.
5. Even though restarting the siphon failed, when a denser beer was added near the end of an initial siphoning, the denser beer did form a small but distinct layer below the less dense beer.

Possible Improvements:

1. In order to allow trapped air to escape fast enough to restart the siphon, the venting needs to happen in a less restrictive way.  The venting also needs to be done in such a way that air is less likely to move into the pipe and stop the siphoning.  To achieve these two goals, the venting hole(s) could be replaced with a venting pipe that runs from the lower section of the drain pipe up through the reservoir to above the fluid fill level.
2. When slicing for printing, enabling a support roof should improve the quality of the reservoir’s interior.
3. The floor of the reservoir could be sloped towards the pipe’s inlet.  This should not only decrease the reservoir volume and improve flow towards the siphon inlet, but also alter the print pattern slightly, which could possibly reduce leakage.

Conclusion:

In this second attempt to 3D print a self starting siphon to construct a Black and Tan, some key principles were successfully demonstrated (e.g., the displacement of a less dense beer by denser beer).  However some aspects of the design prevented the siphon from starting properly when pouring the second beer.  With some additional tweaks, it should be possible to restart the siphon and produce distinct layers of beer with minimal mixing.