I conducted a simple test that consisted of a comparison of the velocities obtained from an AR-15 with the direct impingement gas system with a standard sized gas block, to that of the same barrel with no gas system at all; that is, I completely clamped-off the gas port, turning the AR-15 into a "single-shot" rifle.

Using hand-loaded 69 grain Sierra MatchKings, I fired three 10-shot strings of this ammunition in a row over my Oehler 35-P chronograph positioned 21 feet from the muzzle using a 16” barreled AR-15 with the direct impingement gas system with a standard sized Larue Tactical low-profile gas block.

After firing the three 10-shot strings from the upper using the direct impingement system, I let the barrel cool and then removed the gas tube.  Next, I removed the low profile gas block from the barrel, turned it 180 degrees and reinstalled it on the barrel, thereby completely clamping off the gas port on the barrel.  This gave me a single-shot AR-15 in which I had to manually load and eject each round using the charging handle, giving me the means to determine just how much velocity (if any) is lost due to the auto-loading function of the direct impingement gas system of the AR-15.

Following the same procedure as previously used, I fired three 10-shot strings of the same hand-loaded 69 grain Sierra MatchKings over the Oehler 35-P chronograph positioned 21 feet from the muzzle using the single-shot AR-15.  The results are show in the table below.

As you can see in the table below, there was  a small (but statistically significant) difference of 23 FPS between the grand averages of the velocities from the direct impingement gas system and the single-shot (gas port clamped-off) system.











Atmospheric Conditions

Temperature- 63 degrees F
Humidity - 34%
Barometric pressure – 30.20
Elevation - 960 feet above sea level
Skies – sunny, Jupiter aligned with Mars



....

"Skies – sunny, Jupiter aligned with Mars "

So a little less then 1% velocity loss. What length was the gas system? Carbine?

You could have that much variation in different (lots) of ammo.

Nice work sir!

Thanks for sharing.

Will you please share your load specs for the 69g SMK handload?

I want to start reloading again and was very interested in using this bullet.

Thanks for sharing your knowledge!

That's actually a damned cool thing to know, and the test setup sounds legit to me.

Well done, and thank you!  

I need to try this with my 6.5 tomorrow.  I will just turn the gas off though.

I've been meaning to do this test with my 8.5" 300aac.
I had a batch of subs that must be right on the line because turning off the gas pushed them supersonic (3 position gas block).

Yes.

..

This is the kind of test that you couldn't do with any "AR versus bolt gun" comparison.  By turning off the gas system - without leaving the gas port open to vent to the atmosphere - Molon's shown the actual cost in terms of velocity of the standard direct impingement gas system.

Molon, what sort of changes - if any - did you note in point of impact between gas on and gas off?

This test I have been hoping to see, happy to see it was done by you. Thanks.

Like to ask about velocity variation between groups fired gas on gas off.
Wonder if the gas operating system would make for more velocity spread, opening groups at say 600 yard line.

And what a good build would minimize this velocity spread verse running same gun gasless.

These questions would be better said or explained by you Molon, much respect.

Thanks for all your work.

Very interesting.

Thank you Molon for your work, and for sharing it!

I've read on this subject before and this data confirms exactly what I've read. Right at 20 FPS +\-  very cool write up.

I was far off. I always thought that the velocity loss must be around 100fps, because fully cycling the action requires an relatively great amount of energy, that's what it feels like when you hand cycle it.

Unfortunately, I wasn't "shooting for groups" while I was conducting the chronographing for this test.


..

Too bad.  It looked like a "data two-fer" there.  My hypothesis was "gas 'on or off' will have no measurable effect on point of impact."  Certainly no more than explained by the 20 FPS difference in velocity.

Some vital information is missing....

-What was the extreme spread of the strings?
-What was the standard deviation of the strings?

If the extreme spread was 30 fps, the difference is not really that statistically significant.

Similarly, depending on the standard deviation, the difference may not be all that significant.

Example:

Significant:


Not so significant:

Ideally, the comparison should be between the average, extreme spread, and standard deviation of all 60 rounds fired, and the average, ES, and SD of the two 30 round groups (30 on, 30 off).

If the ES and SD of the two 30 round groups are quite different than the ES and SD of the 60 round group, there is significance.

The insignificant velocity loss is actually significant in that it shows how insignificant the velocity loss is. But I have actually wondered about the velocity loss in gas-operated arms before and am pleased to see how minor it really is. Thanks for going to the trouble to run the test.

That's a very well worded explanation of the difference between "statistically significant" and "significant for decision making".  You can make significant decisions based on the presence or absence of a statistically significant difference between two measurements.  But you have to do the right test - and do them right - to see whether or not that difference exists.

And then run an unpaired, two-tailed t-test that compared all 30 shots fired from the direct impingement gas system, to all 30-shots fired from the "clamped-off gas-port" system.


...

My C-grades in statistics classes can confirm. Nice.

Very cool! Thanks!

So, it's similar to losing about 1/2" of barrel. Unacceptable!

Pretty cool. I don't recall having seen this type of data before.

What was the assumption before the data was obtained? That the DI system robbed some seriously significant velocity?

Interesting that it does not, you'd think it would.

There was no assumption made on my part.  It was just a simple test to see what actual, real-world data showed about the subject matter.



There are those who have made such claims and naturally failed to produce any data of their own to support their claim.



....

It's valid to do an experiment that says "I wonder if there's a difference if I do this..."  These kinds of experiments can tell you whether or not you need to (or want to) investigate further.

...

What stats did you run for this? Do you have a p-value or anything to indicate significance?

As I stated above, I ran "an unpaired, two-tailed t-test that compared all 30 shots fired from the direct impingement gas system, to all 30-shots fired from the 'clamped-off gas-port' system."

The actual p-value and other numbers are long gone from a hard-drive crash.

....

That's a bummer, I was curious to see the numbers.

Still a great test otherwise.

Quick question, and sorry if it was already asked...

Did the gas block actually seal well when tuned around? Most are designed to apply pressure to the side against the gas port, to help it seal.

It probably wouldn't affect things much either way though.

It sealed it well enough that I didn't notice any carbon residue on the barrel at the end of the test.

..

Can someone better explain what an unpaired, two-tailed t-test is?

Unpaired: the data being examined are two samples that are not linked together in pairs.

Two-tailed: the "bell curve" has two tails.  A two-tailed test splits the alpha value between the upper and lower tail.

The unpaired, two-tailed T test compares two sets of data, and allows the comparison of the statistical qualities of the samples, even though those sets of data might be very different in size, magnitude, etc.

Here's a good explanation of this sort of test.

And here's another from NIST.