Interesting article. At least on a basic concept level, the THAI process looks like a major improvement over the conventional schemes, as it actually uses part of the bitumen as fuel rather than more valuable fossil fuels. Furthermore, the more in situ you can make the process, the better.

A major consideration that may not be sufficienly appreciated at this time is the composition and disposition of the combustion gases. Unless these are just allowed to migrate to the surface through the porous overburned, then they would probably have to be collected via some sort of a network of exhaust wells.

The disposition of these gases will not be a trivial matter.The article cites a combustion gas generation rate of 255,000 cubic meters per day for a system producing 660 bpd of bitumen liquid. That is a LOT of gas for a relatively small amount of production.

 Being that the combustion is underground and proceeds from a highly oxidizing state to an oxygen-starved state (i.e, near the face of the liquifying bitumen), it is a certainty that the combustion gases will contain a high fraction of unburned hydrocarbons, possible mixed with carbon monoxide and sulfur dioxide. For a full-scale system I very much doubt that it will be acceptable to release a huge volume of such highly contaminated gases to the atmosphere without some sort of emission controls. A network of exhaust wells connected to a vapor incinerator would handle the unburned hydrocarbons and the CO, but additional controls would be needed to remove the SO2. I'm not saying that air pollution control is going to be a project killer, but it is something that will add to the complexity and cost of a full-scale system.  It is not obvious from the short blurb that I read that this requirement has been fully appreciated.