two small holes at base of barrel

I was wondering on the type of rocket stoves with a barrell over the riser, to radiate heat with the barrel, and the chimney comes out at the base of the barrel.  So air goes into the stove from the opening in front, up the riser, and down in the space between riser and barrel and out the chimney at the base.  

I was wondering if you could keep the front entrance closed, and drill two holes at the base on each side of the barrell to draw air in for the fire, but the front entrance is now always closed.  That way, the fire is still lit and going up the riser and down the area between barrel and riser, but the air now entering must zig-zag through the air that is drawn from the chimney, past the flow of the air to get to the fire.  You have to have two drilled holes to not create a vacuum, but wouldn't that be more efficient, since it's a natural long slipstream that air has to travel all the way to the fire.

I would say that this method may be efficient enough to close the chimney, but leave the two small holes at the base of the two sides of the metal barrel.  The source of the zig-zag is always actually the fire, since the fire zig-zags.  If you assume the chimney is closed and the air has to go to the fire from the two holes opposite each other on the sides of the barrel,  the air that enters (nothing leaves) merges with the zig-zagging that's now present inside the stove, or in other words, there's a tornado in the stove now, and the chimney can be closed.  The air entering through the two holes on the sides of the barrel would have to zig zag to get to the source of the zig zag (the fire), but it would do that, but it now forms a tornado inside the entire stove to allow the air to always be pulled in from the two holes, and nothing ever leaves, so it always sucks through the two holes.

Also, I'm assuming for this to work, the zig-zag (the source), the fire, would have to be de-minimised, so a computer would have to time the rate of closing of both holes in concert, to minimize air and to start the tornado, because the fire is not initially in an optimal state.

Hi Ted;  Welcome to Permies!
Wow , Ted. That sure is a zig zag way of thinking.
If (and that's a BIG if) I follow you correctly. You are wanting to close off the feed tube , after its roaring. Have two small holes in the barrel that somehow get enough air back thru the burn tunnel (while its going) to supply the burn ?
I won't say it won't work because I suspect I'm missing something from your explanation.  
Please clarify for my slow old brain if you would.

What are you hoping for?  Better combustion?  Hotter temps ?
Rocket stoves need vast amounts of air to reach the freaky high temperature's they run at.  Restricting that air may just cause poor combustion  and lower temps... A lot of experimenting went  into the design of rocket stoves.  
They say a rmh that is properly built and run. Has an efficiency of 98% or so.   That's pretty hard to improve on. I would guess that some body has thought of and tried most all variations of rocket science ... I guess no harm in hoping for 99%!

I don't think it's going to work.  I had another idea if the stove were a box, and there was no entrance or exit, and you have a tube like secondary air going into the stove (the air from the tube never touching the inside of the stove, that is, not to give air to the inside of the stove but to draw heat away from the stove), and then an L shape so it exits at the top.  The L-shape would be a chimney, so it would extend for some distance, but the heat of the stove should create air movement in the tube, as well as the draw from it being a chimney.  But then you could make the tube line the entire inside of the stove, so the heat of the stove would cause movements of air in the tube, which is isolated from the inside of the stove.  What I'm trying to say, is to make a pressure difference somehow, so that the air insulation lining the inside of the stove inverts the temperature there, so the stove is cold, and the tube liner draws heat away from the stove.  I noticed on a pressure cooker, submerged in water, right next to the surface it was cold, and any distance away the water was hot in the sink filled with water to cool the pressure cooker.  But my idea is that smoke rings are persistent, so if you can get the stove to burn like smoke rings, it could possibly burn on only the air that was in it to begin with.  And, if you can get it into smoke rings, movement would be smoke snakes that travel through the rings, but the usual way to create a snake is to have two holes, to form a figure 8 around the two holes.  So the stove may not need to be completely closed but have two holes somewhere.  I would also add, that once the snake forms a figure 8, it would have to travel to a center and curl up into a ball before travelling on, but, it would have to be made so that it can keep moving and not disintegrate when it touches the stove walls.

Basically what I'm thinking is that suppose the original idea of two holes is correct near the base of the barrel, and everything else is closed.  Both holes are open.  Air is coming through the two holes, and then a computer closes them simultaneously.  Then air around the  2 closed holes, spins around both holes and a figure 8 forms across the two holes, with the 2 holes of a figure 8 drawing the two small stove holes.  Then it can form a snake, but the problem about forming a snake is that you can never get it to continue forever, as perfect symmetry, the snake will touch the edges (I made a cellular automata that did the same thing) and disintegrate.  So the snake will have to be stabilised by opening one or the other of the holes but not both, because if you open both, a new snake will form destroying the old one.  So probably it would work but it would need a sensor, and a learned neural net, so I'll try to work on my cellular automata, by adding only one dark square at any given time, to see if it will stabilize, but never adding two simultaneously exactly at the same time after the first.  But it also would have to be a square box, as a stove with a hole on the left and a hole on the right, probably.  I think the sensor would be the heat of the outside of the stove, if it's cold it's working, if it's getting hotter, change something.

In the cellular automata I made (white and black squares that are in a 2d grid that appear to blink randomly), it looks like graph paper.  But the pattern I made, didn't have a center.  There were four squares in a box at the center. that were lit in the cellular automata (among other lit squares but the exact center was four squares)  You can see that you can make a pattern like that, whose center is 4 squares.  That was the pattern that I placed some things, and it generated a string or snake that moved around, so I don't think it matters what white or black squares you place on the grid, it's just those 4 squares side-by-side in a box.  So it's possible that you take 4 squares on a graph paper that are next to each other (they are 4 squares in a larger box), and just cut that out in the front of the stove.  There would be a narrow cross in a single hole (square hole), and that would mean a tornado would form (because I think that is somehow significant, and it would automatically generate a tornado that sucks air always.  And obviously the chimney would be capped off, because the air inside the stove would always suck from that thing that is cut out.

If anybody tries to make it, the test would be to put a normal sheet of paper near that entrance in the front of the stove (with the chimney close off), and if when you have a fire in it, it sucks the paper towards the new hole, it should stay there like a vacuum, and not fall.  A regular hole, may suck a paper due to the fire drafting it by chance, but it wouldn't stay stuck there, it would immediately fall.  The reason why it works, is that when you have 4 square holes side-by-side (and up and down ie a square at NW, NE, SE, SW).  I would say the reason why it works is at the smallest scale you have reproduced the asymmetry from the diagonal or hypotenuse of a square.  If on graph paper, every cell can turn on/off randomly, then if you see something that appears to move diagonally it MUST travel faster in the diagonal direction because it has to travel a greater distance (a diagonal of a square is longer than either leg), and all cells update at the same speed.  So if you want to create consistency, in a cellular automata, you wouldn't have a pattern of lit squares with the direct center being just one square, because that wouldn't be consistent with things in large collections being able to travel diagonally faster for apparant motion objects moving diagonally versus direct left-right or direct up-down in television snow.  If you want to be consistent and create STABILITY and not random turbulence is to not have a direct center with one square but a direct center of 4 squares all together in a NW, NE, SE, SW obvious configuration.  Then the smallest scale is not discordant with what happens inside the stove or in a larger cellular automata, and so you would want to have one opening in the stove, in a square shape, with a narrow cross going through the center of that square, (making 4 smaller squares inside the one opening in the usual way).

That's a really neat idea Ted. I've wondered if there was a way to create shaped fires myself. Computer controlled vents do seem like an option. What I'm not grasping in your concept is how the fire flows through the stove shape. Combustion reactions are like whirlpools, an active shape that's constantly changing as the fuel oxidizes into exhaust gases. If you look at pulse jet designs there are some great visuals of how combustion chamber shapes affect the burn.
I'm interested to hear what your goal is in pursuing shaped combustion ideas. I think there might be a way to make a stove that somehow generates some kind of rocket thrust that could power some machinery as well as be a heat source.

I haven't figured it out, but if on a graph paper, you center things so that the direct center is 4 squares, there is no square that is a direct center, so this may be related to aperiodic tiling (quantumfrontiers.com, and scroll down to the tiling area), that in my thing, it may not require larger and larger triangles for instance, if you look at the sierpinski fractal, that is non-repeating because, say you found a repeating horizontal line.  You can always find a larger triangle, so that what you thought was repeating, doesn't include the larger triangle.  If there is no center on a grid of graph paper that's presented, I think it would be easier to find tiles that tile it aperiodically, or non-repeating.  But if the grid has no center, you don't have to make use of a fractal for instance, to get non-repeating, because like in that triangle analogy, you wouldn't have to have things get larger and larger in the layout of the tiles to break up the repeating areas.  I don't see how you could have something repeat if there is no center.  That's the picture at the direct center of a graph paper grid that is not a single square, but a cross or the boundary of 4 squares, because the direct center is 4 squares.

Fascinating ideas, but I'm not clear on how they relate to actual flame path characteristics. Combustion is a complex phenomenon with lots of variables and constraints. You would absolutely have to have a constant supply of fresh air entering the fire, and a similar volume of exhaust leaving; a self-contained fire would use up most of its oxygen quickly and go out.