Usually gasflow in the reduction zone, beginning from the Imbert "expansion cone"
has always been treated as a linear (one direction) flow.
We are accepting that nozzle-beams activate recirculation paths in the oxidation zone.
The same thing happens at the restriction blow beam, straggling but still happens.
When the upper and lower hearth is the same continuous pipe, parted in two by the
the restriction plane, there is an excellent return path for the reduction gases upwards
along the cylinder walls and bending in towards the "restriction beam"...
To be successful, the reduction-cylinder walls have to be doubble-walled, filled with ash, etc.
to keep the return gases at active temperature.
This vertical recirculation of reduction gases in a broad reduction tube
(same tube as for oxidation) keeps the char "process-hot" and instantly ready
for full blast after idle!
This has been a problem for many gassers: Weak responce after traffic lights idling.
(Kristijan, for one, if it is not the mal-precision dispenser)
"Audi Imbert" starts as a "rocket" from traffic lights. No gasoline on board, EVER.