Measuring jet flow with a jug

[moto]

Dear all,

This is a follow-on to a recent thread about replacing or fixing VHB carburetors, https://wildguzzi.com/forum/index.php?topic=112071.0.

In uncharacteristic haste, I ordered some new atomizers for my own carbs during the back-and-forth of that thread, after making a calculation of effects of jet wear that used provided data that turned out to be incorrect. I later wondered whether I had made a mistake in ordering new atomizers, and decided to evaluate the two new ones against the ones in my bike. I'm sharing my results in case they are of interest.

At first I tried to evaluate the jets by timing how long it took for 60ml of water to flow through them. This was inconclusive so I changed the amount to 1000ml (1 liter). I found that this did not work with my equipment (described in a moment), because it produced widely varying values that I decided must come from the weak dribbling at the end of the flow. (Insert your own joke here.) So I came up with a better method that avoided measuring the dribbling at the end.

Equipment:





   A one-gallon jug with the jet mounted in its cap and a hole cut out in the bottom, for filling it with water;
   A one-liter graduated cylinder
   A cell phone with a stopwatch app

Method (all measurements were done very carefully):
   -- Invert the gallon jug, holding a finger over the jet opening, and add 1 liter of water from the graduated cylinder.
   -- Hold the jug over the emptied cylinder (shaking any residual drops of water out of the cylinder first).
   -- Release the water through the jet by removing the finger and start the stopwatch at the same moment.
   -- Observe the water rising in the cylinder and click the stopwatch when the 700ml mark is passed.
   -- Record the reading displayed on the stopwatch, rounding to the nearest second.

Subjects: four #265 VHB atomizer jets, two new and two old.

Data:
       jet   seconds 
  1.     1        78 
  2.     1        76 
  3.     1        77 
  4.     2        78 
  5.     2        79 
  6.     2        79 
  7.     3        70 
  8.     3        69 
  9.     3        69 
 10.     4        79 
 11.     4        78 
 12.     4        76 

The consistency of the measurements within each jet's values was very surprising. Had I expected it, I would have recorded fractions of seconds instead of rounding. The difference between jet #3 and the other four is obvious.

Being a statistician and thus a great fan of belaboring the obvious, I did an analysis of variance, and found very strong evidence that the four needles are not equivalent (p-value rounding to 0.0000).

Jets 1, 2, and 4 took an average of 77.8 seconds to flow 700ml, while jet 3 averaged 69.3 seconds, 11% faster. Shorter durations imply a larger jet aperture, so jet 3 is clearly worn compared to the others, and by a substantial amount. I could calculate the implied diameter of jet 3 on the assumption that the others are 2.65mm, but I will spare us all.

Summary: I developed a reliable method for measuring flow through these jets, using a jug, a cylinder and a stopwatch. Applying the method clearly showed that one jet was not like the others, and the others are very similar. As a result I won't be installing jet #3. Others could use this method to compare jets if needed. (I might use it to check some other jets too.)

Moto

[Rough Edge racing]

 Interesting testing,,,Water has surface tension, try the test with someting added to the water to eliminate surface tension or a different liquid.....Maybe gasoline? 

[Kiwi_Roy]

I used a similar method for measuring the flow through a Griso throttle body only I used a constant head of 500 mm timed to fill a 500 ml container. its very repeatable.
But of course you remember that you were kind enough to take my raw data and make sense of it all.
https://wildguzzi.com/forum/index.php?topic=106043.msg1682048#msg1682048
I'm an Instrument Tech by trade but I could never get Bernoulli's principle to work out.
We used sophisticated software to specify control valves.
I seem to recall that to double the flow through an orifice you have to square the pressure drop.

[jrt]

Interesting testing,,,Water has surface tension, try the test with someting added to the water to eliminate surface tension or a different liquid.....Maybe gasoline? 

or soap, or 50% alcohol.  Easier to clean up

[moto]

I used a similar method for measuring the flow through a Griso throttle body only I used a constant head of 500 mm timed to fill a 500 ml container. its very repeatable.

Yes, Roy, I remember analyzing and posting a graph of your results. I believe my results are even more repeatable (have a smaller coefficient of variation) than yours, though I'd need to look them up. I don't keep a constant head, but I keep the same profile of head across time and stop the measurement before the head becomes too low.

As for the use of water instead of gasoline, Rough, the same argument that I made previously in favor of Roy's use of water instead of air applies here too: the repeatability of results argues I am detecting real differences quite well. It seems  possible that extrapolations from seconds of time to diameters in millimeters might be affected by the choice of liquid, though I haven't yet done that. Surface tension seems less relevant than viscosity in my setup.

Moto

[Rough Edge racing]

  Ok, I understand now.....Mikuni needle jets for example may only have a diameter difference of .0125 mm between sizes. length of the orfice can also effect  flow..Manufacturing tolerance is a factor...Carburetor s are not crude, lol...

[Chuck in Indiana]

Being a statistician and thus a great fan of belaboring the obvious

[Perazzimx14]

The Statisian's blues:

https://www.bing.com/videos/search?q=the+statiscian+song&refig=b30f74aff34044628f5df7d5ff2849ac&ru=%2fsearch%3fq%3dthe%2bstatiscian%2bsong%26form%3dANNTH1%26refig%3db30f74aff34044628f5df7d5ff2849ac&view=detail&mmscn=vwrc&mid=B506A71EA7D42CC4B5F3B506A71EA7D42CC4B5F3&FORM=WRVORC

[radguzzi]

The Statisian's blues:

Too much to think about.  Loves me some Todd Snider.

=====================

So moto,
Interesting process. 

Two questions:
Which other jet was one of the old, worn jets besides the obvious #3..?

How conclusive can the flow rate be with no metering needle in the jet...?

I know that is picking nits however, the flow of the jet alone without the influence of the needle in varying positions surely cannot be said to be a standard, can it...?

Please do not take any offence to my questioning, I am NO mathematician or expert on this matter... obviously.  lol

Thanks, great work here.
Rob



 

[dguzzi]

  Fabric softener (think Downy) may work, its used in printing and paper manufacture for this sort of thing.

[moto]

...
How conclusive can the flow rate be with no metering needle in the jet...?

I know that is picking nits however, the flow of the jet alone without the influence of the needle in varying positions surely cannot be said to be a standard, can it...?

This is a concern I had too, but the Dellorto Guide directly compares flows calculated for jets with and without needles, in section 3.6.5:

In a situation where a very large increase in the main jet size is required, remember that the main jet flow cross-sectional area should not exceed the effective area for fuel flow between the needlejet and the tapered-needle tip. Check this with the following formula:...

The formula given compares the circular area for the main jet with the annular area for the atomizer partly obstructed by the jet needle that I used earlier in the original VHB thread, here: https://wildguzzi.com/forum/index.php?topic=112071.msg1776286#msg1776286
.
 
Moto

[moto]

APPROXIMATE DIAMETER OF THE WORN JET

I will assume the three similar jets are unworn, with diameters Dn = 2.65mm.

The worn jet has an unknown diameter, Dw. Since it flows 11% more fuel than the unworn jet, its area (not diameter) is taken to be 1.11 times that of Dn. That is,

(1)  r * pi (Dn/2)^2 = pi (Dw/2)^2, where r = 1.11.

We now express the unknown diameter of the worn jet as a multiple of the diameter of the unworn jet, that is, Dw = m * Dn. We want to find the numerical value of m.

Substituting for Dw in equation (1):

     r * pi (Dn/2)^2 = pi ((m * Dn)/2)^2
                             = pi [ m^2 * (Dn/2)^2 ]
                             = m^2 * pi (Dn/2)^2

Cancelling pi(Dn/2)^2 from both sides and solving for m:

     r = m^2

     m = |sqrt(r)|
         = |sqrt(1.11)|
         = 1.054

Calculating the implied diameter of the worn jet:

     Dw = m * Dn
           = 1.054(2.65)
           = 2.793mm

So the jet is effectively size 280, rather than 265.

This is an approximation based on my average observed flows (over only 3 observations on the worn jet), the assumption that the three similar jets are 2.65mm in diameter, and the assumption that flows are proportional to areas (which Dellorto, at least, apparently believes). Like Dellorto, I am ignoring Reynolds numbers and the possibility of turbulence, etc., as I make calculations involving small differences in areas. I assume that relative flows of water can stand in for relative flows of gasoline. I do all this in complete and cheerful ignorance of the technical topics just mentioned, placing the onus on experts to show that any of the expected qualms make a important, practical difference.  

I expect to do more measurements, which could lead to a substantial revision of the estimated diameter. This is only a sketch of the assumptions and calculations.

Moto

[moto]

... I believe my results are even more repeatable (have a smaller coefficient of variation) than yours, though I'd need to look them up.

Oops. Wrong. My coefficient of variation is 1.3 percent, based on the 12 measurements I've done. That means the standard deviation around the group means is only 1.3% of the mean, an excellent result.

But your coefficient of variation for the 28 observations in your trails 8-10 (the so-called "baseline runs") was only 0.6%! Amazing.

As I said at the time, hats off to your measurement expertise, Roy! Still, I'm proud to be in the same general ballpark, using a similar water-based approach that benefitted from your example.

Moto

[moto]

...
Which other jet was one of the old, worn jets besides the obvious #3..?

#2 was an old jet, but is statistically indistinguishable from #'s 1 and 4. Something different must have happened to #3 along the way. Probably it became separated from a similarly worn sibling jet, but this was not noticed until now, when I measured it.

Moto