There are a couple of more concepts to get clear on before we move on.
You noticed that we can have “crosswinds”, but realize that we can also have opposing winds of PtA without any more interference than any crosswind would have. In your model, not only might you have H feeding E, but B might be feeding E at the same time. Neither has any affect upon the other as far as delaying unless the max is reached. That max is a flow rate, not a max PtA, so through time, no matter how high the PtA was from any direction, the flow rate dies back down. Of course if you keep feeding more PtA into the model, the max rate will stay high until enough time has passed to make up for how ever much PtA you fed into it. And every bit of PtA that you fed into the model will eventually exit from the opposite side, merely delayed.
The other concept to realize is that not only might E be maxed out, but other points might be maxing out at the same time. Perhaps not only E, but also F and/or G might be maxing out all at the same time. And because of the delay at E, it is more likely that each of the surrounding points are really going to max out also because they could not “empty out” their PtA as fast as normal. Once any one point maxes out, the surrounding points have a higher probability of maxing out as well.
That is how a particle begins to form. The particle is merely the congestion of such PtA getting delayed and causing other nearby points to delay. A cluster of PtA forms that only time could remedy. But because it takes time for the congested PtA to leave that area of space, other PtA has time to enter into that same area. Thus there is PtA coming in and going out from the same area of congestion, the “particle”. When the coming and going are exactly equal, the particle is stable.
Any question about those concepts?