Question about average velocity

Status
Not open for further replies.
spitballer

spitballer

Member
Joined
Dec 20, 2013
Messages
1,050
Location
Central FL
I would like to know if anyone else has had a bullet slow down when the powder charge is increased. This really blew my mind when I saw the figures.

Equipment: .223 bolt action Rem 700, Nosler cases, CCI#400 primers, Nosler 55 grain ballistic tip bullets and Varget propellant.

I tested ten rounds with 26.1 grains of Varget and got the following velocities:
3375
3319
3406
3383
3388
3412
3398
3416
3422
3418 - for an average velocity of 3393.7 fps

The next week I tested 26.3 grains, again starting with a cold barrel, and got:
3216
3356
3368
3355
3346
3344
3352
3386
3369
3365 - for an average velocity of 3345.7

Amazing. Increasing the charge actually decreased velocity. Anyone else have this happen? What do you suppose causes it?
 
What are you using to measure velocity? Hodgdon website shows 24.0 of Varget as a compressed load, how are you fitting in 26.3?

Lastly, what are you using to weight your charges?
 
Last edited:
going over max charges, its said velocity can drop, and thats a sign of rapid pressure increase. I read the high pressure causes the powder to burn faster in a diesel like way, effectively speeding the burn rate, and lowering the velocity. What barrel length are you getting those speed out of?
 
Hodgdon website shows 27.5 of Varget as the max load for a 55 grain jacketed bullet. 24.0 is for a Barnes solid copper bullet. Still compressed, just not the max load for a regular jacketed bullet.
How did you throw the powder? You weigh each charge? Did you start at the minimum and go up or just pick the load?
 
You got huge velocity variations in both chrono series, significantly greater than velocity difference between the two series. In other words, the difference is lost in the greater variation ranges and essentially meaningless.
 
I have seen this a higher charge either results in the same velocity or just a little slower. Only with a bottleneck rifle and a max pressure load. I've also experienced a two times where a cartridge shows pressure signs with a load, but raising the load slightly makes the pressure signs go away. I think the amount of air space in a case and how densely the kernels are packed together in a case affects how the powder burns.

For example I was experimenting with trail boss in a 45-70 lever recently with 350 grain cast. 16 grains filled the case almost to the bottom of the bullet. 17 grains is just barely compressed and gave another 50 fps and tighter extream spreads. 18, 19, 20, and 21 which was the maximum I could get in the case and compress down with the bullet gave exactly the same velocity as 17. 17 grains it is then!
 

Attachments

  • full.jpg
    full.jpg
    130.3 KB · Views: 3
Last edited:
Lots of factors as mentioned up plus

You did not test them on the same day, so perhaps different outside Temp, humidity, alignment of the planets?
 
spitballer said:
I would like to know if anyone else has had a bullet slow down when the powder charge is increased. This really blew my mind when I saw the figures.
Click to expand...

It certainly does happen and is a sign the load shouldn't be pushed any further. If you take it too far, the gun blows up in your face and bullet velocity is not a lot over zero.
 
The chronograph is using the distance from the front sensor to the back sensor to calculate the fps. So if the chronograph is set up a little different from different days that could make the readings some different. If the chronograph is on a slight angle or tilted this could change the reading and I have had problems on cloudy days and different sun angles,weather ,weak batt. they are a lot of variables that could cause your readings to be different. I do try to make the chronograph the same distance and level/straight with the gun.
 
Offfhand said:
You got huge velocity variations in both chrono series, significantly greater than velocity difference between the two series. In other words, the difference is lost in the greater variation ranges and essentially meaningless.
Click to expand...
There's more than 100 fps spread in both groups.

What's the deal with the first shot with the higher charge?
 
...bullet slow down when the powder charge is increased.
Click to expand...

Yes.

mjsdwash in Post #6 explains a mechanism that could account for it. The bottom line is that you are most likely to encounter this phenomenon when you are WAY OVER a safe maximum pressure. Drop back to a lighter load.

offhand in Post #8 notes the large variations in the velocity readings you got. That could be a sign of instrumentation issues or it could be a sign that the powder charges were not consistent. Time to check your reloading processes and procedures.

Sunray in Post #7 asked a very cogent question: "Did you start at the minimum and go up or just pick the load?" If you had begun with a Starting load around 22.8 grains (See Hornady #8) and then worked up, you would have encountered signs telling you that you were entering the territory where added powder did not add velocity long before you actually arrived there.
 
Okay this gives me a better idea what I'm dealing with, from reading the responses it sounds like an unusual situation and my measurements must be off. Judging from this I'll have to do something different - no wonder it took so long to figure this stuff out! Winxb your response is remarkably insightful as usual. Thanks to all members for valued input, and about the unrelated photo - I wasn't smart enough to be able to delete it with editing tools once it got attached and our moderator is welcome to trash it if it's causing confusion.
 
I took a look at your two groups of velocity data and performed a small statistical analysis. There are more variables than just the powder charge, therefore, it is difficult to deduce that the average/median velocity was dropping purely due to the difference powder charge. However, we can still analyse and comment on the data to see if it is indeed significant or not.

Summary
  • There is a significant difference in median velocity between the two powder charges. However, this effect may not be solely attributed to difference in charge mass alone as both sets of data were shot and obtained on different day, possibly under different conditions.
    • 26.1gr = 3,402 fps
    • 26.3gr = 3,356 fps
  • The spread in velocity is not significantly different between the two powder charge groups.
    • 26.1gr = 30.7 fps
    • 26.3gr = 47.3 fps
  • The time series plot shows the order shots were fired, based on the order the data was presented in. Observations:
    • Shot #1 from each powder charge group is significantly lower than rest of data. Likely this is a cold bore or fouling shot.
    • Shot #2 from 26.1gr looks significantly lower velocity than first shot and rest of series. This is likely an outlier, possibly due to throwing a slightly reduced charge.
    • Remaining data shows slight upwards trend from shots 2-10, likely attributed to bore heating up over the course of fire.

Time Series Plot 26.1gr vs 26.3gr.png

BoxPlot 26.1gr vs 26.3gr.png

Data Normality was assessed using the Anderson Darling test and plotted below. A low "AD" value indicates data is likely to be normal in distribution and a P-value greater than 0.050 confirms it is normal.
  • The 26.1gr load produced velocities which are normal in distribution.
  • The 26.3gr load produced velocities which are not normally distributed.
Therefore, we will compare the Median velocity values between the two groups because comparing the mean values required data are normal in distribution.

Probability Plot.png

A Mann-Whitney test was performed to determine if the Median velocity was significantly different between the two groups. Zero does not lie between the two 95% confidence intervals (20.0, 63.0) therefore the Median velocity values are significantly different between the two groups.

A two-variance test was performed to determine if the standard deviation, or spread, in the velocity is significantly different between the two groups. We deduced that one set of data were not normal, therefore we will observe Levene's test as the F Test is only for normal data. The test produced a P-value of 0.896, which is greater than 0.050, and therefore confirming that there is no significant difference in the spread in the velocities between the two data sets.

T-Test and Variance.png
 
WelshShooter said:
A two-variance test was performed to determine if the standard deviation, or spread, in the velocity is significantly different between the two groups. We deduced that one set of data were not normal, therefore we will observe Levene's test as the F Test is only for normal data. The test produced a P-value of 0.896, which is greater than 0.050, and therefore confirming that there is no significant difference in the spread in the velocities between the two data sets.

View attachment 766947
Click to expand...

Excellent analysis Welsh Shooter!

Something that Spitballer did not do was fire both samples on the same day. Therefore there is the real possibility of instrumentation error. I regularly shoot a "reference load" over the chronograph prior to testing to determine if the chronograph is properly aligned. Small variations in alignment create variations in velocity readings. Temperature changes, primer changes, etc, etc, all create differences in velocity. And even then, the velocities averages are never exact, because we are dealing with small sample sizes that have wide standard deviations. Remember, one standard deviation represents 68% of the normal curve. http://www.oswego.edu/~srp/stats/z.htm I don't know the sample size it takes to get convergence, but it is pretty large. So for small sample size you have to resort to the statistical tools that Welsh Shooter used to determine if the populations are in fact different.
 
Thank you Slamfire, and you are correct. I performed the analysis in a software called Minitab which also contains a feature that relates Sample Size, Measured Difference and Power. So long as you enter two of the three aforementioned data points along with a standard deviation, the tool will fill in the gap.

In the OP's example we have a sample size of 10 and a standard deviation of 30 and 47 respectively. We want to have 95% confidence detecting a difference in the mean values (standard for these types of tests). The test confirms that it is capable of determining whether the averages are significantly different or not so long as the difference is greater than 80 fps.

Power Curve.png


If you want that difference of detection to be smaller, per the OP's example this measured delta is 48 fps, you would need a sample size of 26 shots per group.


Power Curve 2.png



Sorry OP, a little bit off-track but I do enjoy stats me :)
 
Status
Not open for further replies.
Back
Top