I use my own
RMH to heat a poorly insulated space, but these are the impressions I got from using spaces around
Wheaton Labs and watching Paul use his in the FPH. You don't run it at partial throttle unless you are trying to keep the aesthetics of the rocket roar going. Burn at 100% until you have generated enough heat to charge the mass, then let it burn out and cover the
feed. This gives the most efficient burn and lets you change focus away to other tasks.
I'd say this would be filling the feed 80-100% with wrist sized sticks every 25-40 minutes. The system can be turned up to 11 by using matchstick sized
wood or straightening out sticks as they burn down and fall if you care to. I run mine lighter like it looks like you are, but am trying to keep constant heat radiating from the barrel and the mass is just to extract some more heat before exit. Using larger chunks of wood is good for throttling down the burn also, but wrist sized is a good average size for a quick, hot burn since smaller is much more finicky.
As for design issues, I concur with what has been posted about the barrel. 50mm top gap was the minimum mentioned by Ernie and Erica in the book, but ongoing experimentation has shown the 80mm provides consistent results while still allowing enough heat on the surface of the barrel to use it in a stove fashion. If you don't want to cook or heat drinks on it, it sounds like it can be as high as you wish.
Pushing the minimums here seems the likely source of poor performance. Given the emphasis of Ernie and Erica and their mentor Ianto on affordability of construction, I'm guessing minimums in the book are based around common, thinner barrels. Stainless steel is thicker and does not conduct heat from inner gases to the radiant surface as readily from my understanding, so running on minimum CSA there could hamper your thermal pump's 'push'. Also, did you use the actual measured inner diameter of the barrel to calculate that CSA? The picture of the new riser just looks like it has next to no space to where the barrel sits.
Finally, I was going to mention the cleanout Ts. The length of the dead ends seems to be about twice normal, even more for the one by the chimney. I was thinking in terms of diminishing pressure to guide gas through the turn, but Scott sounds like he has some
experience with such phenomena so maybe turbulence from 'bounce' is a more accurate description. Intuitively a closer flat surface at the turn would seem likely to produce greater bounce to me. It would be an interesting experiment if fabricating plugs to try a 45* angle one that would fill the space like the wall of a regular outside of the 90* bend's duct...