Wednesday, November 29, 2006

AV - Flow-Bench vs Reality

--- In, CaptonZap asked:

> Have you ever heard of a flow bench that pulses? -----------------------------------------------------

Some fluidics applications -- gates and the like -- are tested using a flow-bench that emulates real-life situations. But I've never heard of such being applied to an automotive flow-bench.


>I don't know if the difference in > flow would warrant the trouble to make a pulser for the bench. Any thoughts?


The original purpose of the flow bench for head work was to provide a measure of merit rather than quantified data. But it wasn't long before we saw flow-bench figures converted into purely notional units of cubic feet per minute (what you actually measure is the pressure differential) and used to sell junk to the Kiddie Trade. The same sort of 'bigger is better' hype is used to sell cams, valves, cranks, jugs and so-forth.

After my second tour in Vietnam I built a rather elaborate flow-bench. Over a period of about seven years the most valuable thing it taught me was how little I knew. And I'm not just talking fluid dyamics :-)

As you've already guessed, measuring the flow of air can only provide a measure of merit -- this head flows more (or less) than that head. But fuel-air does not behave the same as air alone. Smoke or other benign suspended colloid, in the proper ratio with air, provides a more realistic test -- and often produces results significantly different than when using air alone.

The idea that a simple measure of merit is good enough for selecting the 'best' heads is only valid for dragsters, where fame & fortune is based on a run-time of six to ten seconds. As soon as you add durability to the equation -- and measure it in hundreds of hours rather than tens of seconds -- you may as well throw away that huge pile of heads you've been working with and start all over. If you do that, don't be surprised if your best efforts looks remarkably similar to the heads off existing aircraft engines :-)

An even tougher test is to discover that after a thousand hours of work you've managed to confirm what Sir Harry Ricardo defined in his book... in 1920.


Lemme give you an example of why the flow-rate thing is mostly hype. Let's use a 2180cc engine having a volumetric efficiency of 100% (yeah, I know -- impossible -- but work with me here). 2180cc times 0.06102 equals 133.0236 cubic inches. But were talking a four-cylinder Otto cycle engine so the actual displacement per revolution is just half that or about 66.5 cubic inches per rotation of the crankshaft.

Now pick an rpm, something seriously silly for an airplane, such as 3200... because you want to make the prop horribly inefficient and wear out the engine quick like a bunny. Or whatever :-)

66.5 cubic inches times 3200 rpm is 212,800 cubic inches. Divide that by 1728 and you've got 123 cubic feet per minute.

That's a totally fallacious figure because at that rpm your VE won't be anywhere near 100% but lets use it as a bench mark... or even round it up to 125 cfm just to make the figuring a tad easier.

Keep in mind, that's the also flow through the carb for the entire engine -- all four jugs.

Now take a look at all those trick heads they sell to the kiddies. Huge valves... which run hot and seal worse than smaller valves... but a nice match for the huge ports and cut-away valve guide boss... that gives you flow rates per cylinder as high as 350 cfm. Wow! Gotta be good, eh?

If one jug can flow 350cfm, with four-cylinder Otto cycle engine that means it has the potential to pump 700 cubic feet of air per minute! Gosh! Is that great, or what?

Did anyone notice that the maximum amount of air the engine can use is only 125 cfm? Apparently not, because hi-flow heads sell like hot-cakes :-)

Wanna build an engine for a replicar like the Beck 550 Spyder? Then you want something that can spin seven grand and hit one-forty on the straight-aways. That's when you go for hi-flow heads, six pound flywheel, titanium valves and four 40mm Dellortos.

But not to sling a prop.

The airframe dictates the prop and the prop dictates the engine. That's when you discover that the actual flow-rate of your engine is something less than 90cfm and that even Single-Port heads can give you that. Of course, you put your stock SP heads on the flow-bench and clean up the ports and swirl-polish the valves and attend to a host of other minor chores and come up with a marked improvement in your volumetric efficiency for nothin more than a bit of labor.

Why the interest in volumetric efficiency? Because in a normally aspirated engine peak VE tends to coencide with peak torque. And despite what the hucksters keep yelling, with a fixed-pitched prop mounted directly to the crankshaft the key factor in producing thrust is how much torque you have available at that particular rpm.


Which prolly isn't the answer you expected, although you did ask for 'any thoughts' :-)


No comments: