[Madact]

Originally Posted by Nigel_S

My understanding is that you are a fair bit better off with long straights and fairly sharp 90 degree bends. If nothing else, it is easier to get the correct length since the flow will not go around the centre line of all those varying angle bends, but I think it is better for maintaining the gas speed too.

Found this reference Bends, flows and pressure drop in single phase flow

It's for steady flow though, of course, but it implies that, for a steady-state flow at least, wider curves are always better than narrower ones if you have the length to spare (and boy, do I have the length to spare...). Also, looking at the bend loss coefficient graph, it looks like "once you start a turn you may as well keep going" - the coefficient for 180 degrees, for example, is always less than twice the coefficient for 90 degrees. If you *need* to turn 180 degrees, a 180 degree bend is definitely better than 2 90 degree bends with a straight section, if the pipe length in both scenarios is equal... this implies that a continuous ram horn bend would also be better than a 'trumpet' style arrangement.



Also, odd mathematical thing, but for a given angle of bend, the difference between the length of the inner edge and the length of the outer edge is a constant regardless how sharp it is... consider a circle of diameter D (the inner edge) and a circle of diameter D+Pd (the outer edge - Pd is the pipe diameter). The difference in circumference is pi*(D+Pd) - pi*D = pi*(D+Pd-D) = pi*Pd.

Considering the limits, an "instant" 90 degree bend would intuitively be a worst-case scenario, regardless of whether we're considering pulses or steady flow, while a sufficiently wide bend is indistinguishable from straight pipe. Again intuitively, if we increase the radius starting from zero, things will get better for a bit. Now, it's possible that
(a) there's a region where things improve as radius increases
(b) if (a) is true, there may be a radius that flows [i]better[i] than straight
but (b) seems implausible and (a) seems unlikely given we know what happens in the limits. Also, if (a) were true, there would be a 'local minima' ideal radius somewhere that was better than a radius which was a bit smaller or a bit bigger, if that were the case I suspect it would get a notable mention in physics classes and fluid dynamic texts

Could be wrong about this, and dynamic behaviour of pulses is complex enough to possibly invalidate some piece of reasoning here, but I think it's correct.

Now, there *is* a tradeoff in the steady state here as per the graph if you're trying to optimise for least resistance and don't care about length - if you can save some pipe by doing smaller radius bends, it's probably worth it, though from the graph, the region where bend radius < 1.5D is a region you don't want to be.