Optimal TEG
There are many ways to set up a TEG; and throughout SS14 history the TEG's power has fluctuated significantly.
But regardless of what its strength has been - there has always been one optimal way to set up a TEG (that is legal on all servers; IE does not utilize subframe - though you can make a subframe variant of the same core idea and it works just as well).
This is what I call a Two-Stage Burn. IE - you burn in two stages; the first stage focuses purely on maximizing Tritium production (IE maintaining close to 1650 Kelvin; or 700 Kelvin on Frontier- and otherwise maximizing cook molage within whatever container you're utilizing for this stage). For Tritium production, you want as few waste gases present as possible to make more space for plasma and oxygen - and you want the ratio inside the chamber to remain as close to 0.986% Plasma as possible (3% on Frontier).
This will maximize the total amount of Tritium produced - and since well over 95% of the total thermal energy from a burn is in the Tritium, this therefore also maximizes energy production.
However, Tritium itself desires dramatically different burn conditions than Plasma - it wants 2/3rds oxygen and 1/3rd Tritium to maximize its burn rate and for a TEG you want the temperature to be as high as possible to maximize the amount of energy you can move and therefore maximize the output of the TEG (which is not optimal for producing tritium; since higher temperatures = less molage within identical pressure/volume constraints).
Thus, the Two-Stage burn. Stage 1 makes Tritium, Stage 2 burns the Tritium. For added efficiency and to minimize the odds of burn rate slowing due to clogging, you can additionally add a third "Stage" which is simply storage of the water vapor. This way there is more room in your Second stage for Tritium burning, and thus a much higher maximum possible operating temperature for the same given throughrate of Tritium- or simply a higher throughrate.
Doing this with burn chambers is of course very possible; though the Tritium burn chamber will likely be relatively small since Tritium burns very compactly compared to Plasma. Most stations also aren't set up for this kind of system - and so you'll end up spending alot of time just building it. If you know subframes and your server does not Forbid them, then you can do all of this inside just 1 (or 2) Subframes and be more or less fine - though if your server does not have heat pumps you will need some way of transferring the heat out of the subframe.
I provide no specific examples of setups here since it can take a myriad of forms - and its the concept that is most important. Results will vary depending on the local conditions of the TEG and Tritium fire reactions - for Funky, which is the server this forum is based in, using two 100mpa gated subframes for tritium production and burning and then storing the vapor in the often provided TEG chamber is an ideal setup that will net (if running perfectly) roughly 11 Megawatts, at the low low cost of 30 each of plasma and oxygen per second. This produces in a PTP (Power Transmission Point) about 40 speso per second, which is a horribly horribly inefficient exchange rate compared to simply producing Frezon.
So on Funky, do not use the TEG - pretty much ever.
A Turbine However? Well... you'll need 9 of them. The reason the TEG produces 11MW is because it converts only 1/8th of the total thermal energy - a Turbine converts 100%. Meaning with Turbine conversion you can do 90 Megawatts; and since Turbines cap at 10MW, that means you'll need 9 turbines to sustain a capped subframe based two-stage burn. Though with the Forky move, the Turbine becomes uncapped - so let your two-stage burn dreams run wild!
Or don't - because even at 90mw the exchange rate on the PTP for this power is laughably bad compared to simply producing Frezon. The conversion rate, btw, is ~.75 speso per oxygen and 6 speso per Plasma.
When used even at 100% efficiency for Thermal energy, in a PTP the 30/second Two-Stage burn yields only 83 Speso - compared to the 202.5 speso value of the gases consumed in Frezon alone (to say nothing of their potential in even higher yield gases, such as Hypernob or HFR-based recipes)