Venturi Test

LAST EDITED ON Apr-19-04 AT 01:59 AM (CDT)
 
Preston and I did a pressure test today on the venturi that I wrote about on a previous thread. I don't know how to post text amongst pictures, so I guess I'll post this in multiple messages.

Background:
We set out to see if we could make a venturi for an OS .15 such that the venturi had a substantial pressure drop without much engine power loss. I then did some calculating and decided it couldn't be done. Folks said that I was neglecting that air goes through a venturi in bursts, so the average pressure drop would be more than it would be if air flowed through the venturi at a constant speed. I decided to do an experiment to see how the pressure varied at the throat of this venturi.

Test Apparatus:
The first two pictures are of the venturi mandrel. I cooked some carbon-epoxy prepreg cloth on this to make a venturi that's about 5/32" min. diameter and that diverges with the wall at 7 degrees from the centerline. The third picture is the venturi and electronics mounted on the engine on the test plane. We made a little circuit with a MEMS pressure sensor and a quad op amp to buffer and amplify the pressure signal. No fuel goes through the venturi. It gets squirted directly into the crankcase for this experiment.

Ready to test:

Ron B.
F4Fguy

We're all agog!!When's the next installment?

The first oscillosope trace is with the engine not running. The wide line is because there's some 120kHz noise. I don't think the noise is coming from the test circuitry. Scale is .05 volts per big vertical division and 2 milliseconds per big horizontal division. Note that there's about a .05 volt bias at zero pressure. The second trace is with the engine running. Mean venturi pressure drop is about .48 psi. Amplitude of the variation is about .07 psi. Note the interesting transient, which seems to ring throughout the period.

>>>Note the interesting transient, which seems to ring throughout the period.<<<

Really nice! If I calculate properly then the rpm was ~14 000 and its period is ~1 ms. The venturi is open pipe 1/4 wave resonator what well responds to 30mm long pipe (if I calculate little less then sound speed because of lower average pressure). Is it properly interpreted?

If yes, does it mean that the crank closes relatively high intake speed at those 14000 rpm – because the first peak is to high pressure. It either means that the timing is too slow for those 14 000rpm, or our cranks are closing too early at all? Or that is what makes OS "stunt friendly"?

I think it was about 12,000 RPM.

Yes, I see, it is 12000. So it means the shaft timing is low even for those 12000 right?

LAST EDITED ON Apr-19-04 AT 01:54 PM (CDT)
 
> If I calculate properly then the rpm was ~14 000 and its period is ~1 ms.

The period is about 0,005 seconds (each horizontal division is 2 milliseconds and we see 2 cycles taking about 5 divisions), so about 12K/minute

AND what's the point?
I've been shipped wrecked

Scott(Not FRIDAY)Riese

>AND what's the point?

Howard's first post sets the stage:
> We set out to see if we could make a venturi for an OS .15
> such that the venturi had a substantial pressure drop without
> much engine power loss. I then did some calculating and decided
> it couldn't be done. Folks said that I was neglecting that
> air goes through a venturi in bursts, so the average pressure
> drop would be more than it would be if air flowed through the
> venturi at a constant speed. I decided to do an experiment to
> see how the pressure varied at the throat of this venturi.

The experiment suggests that folks were wrong. Instead, the pressure is relatively constant, varying about 15% across the cycle.

> Mean venturi pressure drop is about .48 psi.
> Amplitude of the variation is about .07 psi.

So perhaps Howard's thought, namely "it can't be done", is accurate.

I would expect the pressure to vary more in shorter venturis. This one is longer than usual. I would also expect shorter venturis to be less efficient than this one.

its not clear from the pictures where the pressure is being measured. At the base of the venturi, just above the shaft opening?

How about a test of a conventional cross bar needle valve setup like the stock one?

How does the power of the engine compare between a stock venturi and this one?

Pressure is measured at the venturi throat (or maybe a tad downstream) by a hypodermic-needle tube sticking about 1/16" into the venturi.

Other tests could be done, but I gotta get some airplanes flying.

There might be a problem with the test setup. The air chamber in the pressure transducer and the tubing leading to it would act as a low pass filter. This would tend to mask the rapid changes in pressure.

Pat MacKenzie

> There might be a problem with the test setup.

Possible.

> The air chamber in the pressure transducer and the tubing leading to
>it would act as a low pass filter. This would tend to mask
>the rapid changes in pressure.

The spec of 100 microseconds (see below) would include the air in the chamber.
And, it you'll very closely at the 3rd picture (and use your imagination),
you'll see that the tubing is very short, as short as we could make it.
That's why the sensor is mounted right next to the venturi.

Still, there's always room for error. In traditional fields,
other scientists would apply to the NSF for funding to replicate
our results. Don't delay, apply today!

Maybe so. I kinda expected that. We could have repeated the test with a bigger tube to see if the tube was filtering the data. We could have calculated that, too, I guess. So the .07 psi amplitude is a lower bound. If the high-frequency transient really is coming from the venturi airflow or acoustics, then the frequency response of the kit we used is adequate to measure the amplitude of the fundamental of the once/revolution pressure variation. The transient may be coming from the circuit board banging against the needle valve. I don't really know. If we get a heap of DARPA money to look further, we will.

Pat, I think if the unit can measure at that transient frequency, it must be quick enough to measure at rpm frequency with enough precision.

However I would really try to use longer tubing to be sure that the transient is in venturi and not in the sensor tubing … because now it looks the venturi length and the sensor tubing is of the same length and thus can have the same natural frequency, and thus it is difficult to distinct what is what. But anyway, it does not change situation too much, if the transient effect is in sensor tubing then its frequency is high enough to be sure it is enough for measuring at RPM frequency.

>The first oscillosope trace is with the engine not running.
>The wide line is because there's some 120kHz noise. I don't
>think the noise is coming from the test circuitry. Scale is
>.05 volts per big vertical division and 2 milliseconds per
>big horizontal division. Note that there's about a .05 volt
>bias at zero pressure. The second trace is with the engine
>running. Mean venturi pressure drop is about .48 psi.
>Amplitude of the variation is about .07 psi. Note the
>interesting transient, which seems to ring throughout the
>period.

Ron B.
F4Fguy

Howard:

A couple of points:(1).46PSI IS significant suction.(2)You do show pulsation(and apparently some resonance).(3)What is the peak to peak response time of the pressure sensor?It's extremely difficult to measure transients with any accuracy in low pressure air.That's one of the reasons for recommending the hypo tubing.

> (2)You do show pulsation(and apparently some resonance).

Sure. The point of the test was to look at the magnitude of the pulses (versus guess about them).

> (3)What is the peak to peak response time of the pressure sensor?

100 microseconds.

The pressure drop was five times the biggest number I expected for a no-loss venturi. My tentative explanation is that the efficiency is low; the engine is just sucking through a flow restrictor. A simple test would be to pull the venturi out and see if the engine runs faster. I guess we should also measure the pressure next to the crankshaft with and without the venturi, but that might have to wait for the DARPA money.

The simplest way to measure efficiency is to compare pressure before and after the narrowest place in venturi at the same diameter – for example at place where is 2x bigger diameter that should be easy and enough (1/4 of velocity at quadratic response).

The amount of pressure difference shows amount of lost kinetic energy in tube. If the efficiency is 100%, means no lost kinetic energy and no dynamic friction. The pressure must be by Bernoulli equation equivalent.

The pressure difference is lost kinetic energy and back by Bernoulli can be calculated amount of “pressure” energy in stream before and after and thus efficiency.

>The simplest way to measure efficiency is to compare
>pressure before and after the narrowest place in venturi at
>the same diameter – for example at place where is 2x bigger
>diameter that should be easy and enough (1/4 of velocity at
>quadratic response).

OK, I see. Good idea. Hard to tap the venturi that accurately, though.
Especially on the intake side. Maybe if we make a new mandrel
and make provision for the tap while the mandrel is still on the lathe?
Hmmm, hmmm, hmmm.

Howard's idea is easier. Just run it with and without the venturi, or with the test venturi and the standard OS venturi/NVA.

The whole point of that style venturi is to generate a large suction for fuel draw but let the air pressure recover at the shaft valve so you don't lose any power. Just the fact that it generates significant pressure drop in the venturi doesn't mean it is restricting the engine.

>Howard's idea is easier. Just run it with and without the
>venturi, or with the test venturi and the standard OS
>venturi/NVA.
>
>The whole point of that style venturi is to generate a large
>suction for fuel draw but let the air pressure recover at
>the shaft valve so you don't lose any power. Just the fact
>that it generates significant pressure drop in the venturi
>doesn't mean it is restricting the engine.

Ron B.
F4Fguy

Phil:

Exactly.