Jane Ashworth wrote:Would it be better to position the channel within the portal with the air ejecting from the top? And do you recommend having a flap here?
Your questions show you didn't get the principles (yet). So I'll try explain it again, although it would be better to stop thinking and just do what is recommended. Air ejecting from the top inside the port itself wouldn't work any better, rather the other way around. Just because the incoming air wouldn't be heated up as much.
The basics are on the website but here we go again: the port act as a temporarily restriction in the smoke path. As such, it is a venturi wherein the gas speed increases and the pressure decreases. It's a law of physics called the Bernoulli theorem, part of the law of conservation of energy. It doesn't matter whether or not you understand how it works, even believing otherwise won't change laws of physics.
Now how to utilize this phenomenon in order to get a sufficient quantity of fresh, hot air into the port. In order to understand this, it is imperative to realize that the lowest pressure in the port happens to be just a fraction of an inch beyond the narrowest point, seen from the firebox' side.
By hanging down a small piece of the channel over the opening, that said narrowest part of the port is the p-channel itself. So the lowest pressure and highest velocity is generated where the air from the channel comes in. That way, air is literally sucked in.
What has been done wrong in your implementation is just a flap hanging over the port, with generous openings left and right. This is so much cross section area together that the narrowest spot in the port isn't where it should be but further downstream instead. That flap should be a U-shape, and the channel should be at least as wide as the port. Entering from the side doesn't help either, none of my drawings shows that and for a reason. The part of the channel inside the firebox shouldn't be horizontally but vertically instead for best performance.
To conclude: please see that the recommended dimensions, shapes, sizes, proportions, end temperatures and whatnot are adhered to. Just copy the whole of the core design and it'll work, right out of the box. For reference and pictures: see http://batchrocket.eu/en/designs#pchannel
. Both of the embedded pictures show how the p-channel should look like. If at all possible, build just the core outside using bricks and clay/sand to seal the crevices and fire the thing up.
P.S.: the bench together with the water barrel is way too much. As long as it's there the thing won't work as desired so measure the temperature inside the exhaust pipe to the roof, not the outside.