Here's a new reactor insert for the GEK we've been discussing internally around the shop for some time.  As recent discussion on the forum has rightly wandered into ideas for better zone separation, internal tar recycling, and open combustion chambers separate of the char bed, I thought I should surface this.  This is where I think you end up when you try to integrate these features into a single reactor, while maintaining a concern for ease of building.  

The design has been named the "Triple Hips" given the three horizontal restiction plates used to formally separate the four zones. The restriction plates also create open annular spaces where gasses can be handled separate of the packed bed.  These spaces are used to mine tar gas from the pyrolysis zone, then direct it to an open swirl combustion chamber where tar combustion and cracking happens separate of the char bed.

Likely the most important feature of this design is the dedicated combustion chamber removed from the char bed.  Combustion in a mixed gas/char bed is always fighting against the endothermic potential of the char, as well as the ever changing shape of the void spaces between the char.  If the goal is to combust some tar gas to create high temps, then maintain and distribute those high temps so as to crack the excess tar gas, a char bed is a very non-ideal place to do so.  A glowing char bed is somewhat like a cold water bath on the products of combustion-- CO2 and H2O.  The moment oxygen is depleted in combustion, reduction of the "flue gasses" starts over the char, with very quick drops in temperature.  This happens with a highly variable rate and location given air distribution and fuel void space particulars. 

The result is the well known difficulty of maintaining a full plane of combustion across the char bed between the nozzles.  Vary gas flow rates, or fuel size and shape particulars, and things fall apart quickly.   Thus a packed bed downdraft is always plagued by a fairly narrow "sweet spot" of fuel type and pull rate where the full hearth is enveloped in the proper temps.  Moving tar gas combustion and cracking to an open combustion space allows for these processes to proceed to completion without depdendencies on the char particulars.  The combustion space and duration is determined by the particulars of the built combustion cup, not the ever changing particulars of void space in the packed bed. 

The Triple Hips reactor uses ejector venturis to purpose scavenge tar from the top of the pyrolysis zone and pull it down into the swirl combustor.  This active tar scavenging also creates an upward flow of hot gasses to the ejector venturi inlet tubes, thus increasing upward the length of the pyrolysis zone.  Extending pyrolysis upward increases pyrolysis residence times, while reducing overall pyrolysis temps.  The result is more complete pyrolysis before reduction, and the tars produced at lower temps will be easier to crack.

Pyrolysis in a typical nozzle downdraft is only a couple inches deep over the nozzles, resulting in short residence times at very high temps.  The high temp produces more complex refractory tars that are more difficult to crack.  The short residence time often means larger fuel chunks do not finish pyrolysis before falling into the combustion and reduction zones.

In addition to the upward extended pyrolysis zone, the annular swirl combustor around the char bed creates a second oppotunity for tar mining and cracking.  The center tube through which the char falls down into reduction is made of perforated stainless steel.  Any tar gasses that make it to this low point in the packed bed should be pulled outward into the surrounding swirl combustor.  The drawing above shows my current guess as to the needed length for the swirl combustor and secondary tar scavenging dynamics.  We can easily lengthen or shorten it as we learn its behanior in operation. 

As the design is not relying on air blast bed penetration and combustion for tar cracking, we should be able to maintain wider fuel flow spaces, and thus enjoy reduced bridging problems.  Still, it will likely be a good idea to put a stirrer rod down the center.