Effects of Case Weight Variation on Accuracy

[Bc'z]

Well hell!!!
After reading this very informative thread, it looks like I'll be weight sorting cases.

 

[Texas10]

I have a question for the group. Has anybody taken 2 extreme weight cases from a single date code, single vendor and milled them in half so you could see the wall thickness as well as the case head dimensions and see where the weight changes are occurring from?

If that has been done what has it shown?

David

We're talking tenths of a grain. Imagine trying to figure out where that tiny bit of brass is hiding. It may not be INSIDE the case at all, it may be the way the head and extractor notch is cut. That's why there is no definite relationship between internal capacity and weight, not to mention the fact that slight differences in case length due to expansion after firing would lead to slight differences in internal volume with no weight change.

 

[Bc'z]

We're talking tenths of a grain. Imagine trying to figure out where that tiny bit of brass is hiding. It may not be INSIDE the case at all, it may be the way the head and extractor notch is cut. That's why there is no definite relationship between internal capacity and weight, not to mention the fact that slight differences in case length due to expansion after firing would lead to slight differences in internal volume with no weight change.

This has always been my thoughts. But...
Wouldn't base to datum measurement being consistant on sized brass that has also been trimmed to length help in making a good volume comparison.

 

[Texas10]

This has always been my thoughts. But...
Wouldn't base to datum measurement being consistent on sized brass that has also been trimmed to length help in making a good volume comparison.

Yes, I think you get closer to being consistent in internal volume, but again, weight can change a lot in the area behind the web and yet not change the internal volume.

One other consideration is base to datum (shoulder) and shoulder to mouth, since if you bump the shoulder a bit further forward, but trim the case referencing the base, you get more internal volume without a change in OAL. And vice versa.

I have weight sorted brass, and done a H20 sort. They do tend to follow one another, but it's only a tendency. There is no direct correlation IMOP.

But if you've found a node that is not sensitive to slight changes in propellant charge weight, something we strive to do, then little differences in internal volume should not matter.

As has been said many times, let the target tell the story.

 

[Bc'z]

Yes, I think you get closer to being consistent in internal volume, but again, weight can change a lot in the area behind the web and yet not change the internal volume.

I have weight sorted brass, and done a H20 sort. They do tend to follow one another, but it's only a tendency. There is no direct correlation IMOP.

But if you've found a node that is not sensitive to slight changes in propellant charge weight, something we strive to do, then little differences in internal volume should not matter.

As has been said many times, let the target tell the story.

Thanks for strengthening my thoughts.
I've been replacing brass for our bolt guns with fresh batches vs fired factory ammo for the brass, with unknown firings/lots
I've started sorting bullets and see that improvement.
Was thinking seriously about sorting cases and not looking to it.

 

[Ned Ludd]

We're talking tenths of a grain. Imagine trying to figure out where that tiny bit of brass is hiding. It may not be INSIDE the case at all, it may be the way the head and extractor notch is cut. That's why there is no definite relationship between internal capacity and weight, not to mention the fact that slight differences in case length due to expansion after firing would lead to slight differences in internal volume with no weight change.

This is not correct. There IS a general linear relationship between case weight and case volume within a single Lot# of brass. I've posted rock solid evidence that this is true on several occasions recently. Anyone is free to choose not to believe that if they wish, but that doesn't change that fact. Edited to add: you will always find a few outliers that do not fall exactly on the case weight versus case volume trend line. This does not change the fact that the linear relationship/trend exists.

The external dimensions of fired cases are extremely uniform, possibly even more so than cases that have been re-sized. The only significant source of weight variance that would not affect internal volume is in the extractor groove and the primer pocket. The extractor grooves are are machined with very uniform tolerances. The primer pockets are also very uniform in both diameter and depth. If they were not, primers would not fit the same from case to case, and primer pocket uniforming tools simply wouldn't work. Neither of those things happen, indicating the variance between primer pockets and extractor grooves is not large enough to affect the relationship between case weight versus case volume in a significant way.

On top of that, the extractor groove represents only a very small fraction of the internal volume of a case, anyhow. Any variance in the extractor groove volume between cases therefore represents an extremely small fraction of the total case volume. Whether someone believes sorting cases by weight as a surrogate to determining actual case volume is a worthwhile endeavor is is totally up to the individual. However, claiming that there is no relationship between case weight and case volume is demonstrably false and sorting cases by weight will generate more consistent internal volume than by doing nothing at all.

 

[Bc'z]

This is not correct. There IS a linear relationship between case weight and case volume within a single Lot# of brass. I've posted rock solid evidence that this is true on several occasions recently. Anyone is free to choose not to believe that if they wish, but that doesn't change that fact.

The external dimensions of fired cases are extremely uniform, possibly even more so than cases that have been re-sized. The only significant source of weight variance that would not affect internal volume is in the extractor groove and the primer pocket. The extractor grooves are are machined with very uniform tolerances. The primer pockets are also very uniform in both diameter and depth. If they were not, primers would not fit the same from case to case, and primer pocket uniforming tools simply wouldn't work. Neither of those things happen, indicating the variance between primer pockets and extractor grooves is not large enough to affect the relationship between case weight versus case volume in a significant way.

On top of that, the extractor groove represents only a very small fraction of the internal volume of a case, anyhow. Any variance in the extractor groove volume between cases therefore represents an extremely small fraction of the total case volume. Whether someone believes sorting cases by weight as a surrogate to determining actual case volume is a worthwhile endeavor is is totally up to the individual. However, claiming that there is no relationship between case weight and case volume is demonstrably false and sorting cases by weight will generate more consistent internal volume than by doing nothing at all.

I've read your posts on this subject which has me thinking strongly about weight sorting.
The part that's most dreadful is the thought of measuring volume, air bubbles n all leading to inconsistencies on my behalf.
Or is weight sorting good enough?
I did grab 10 pieces of fresh lapua brass in 30 06 and only had .9 variance.
Above post not sure who without re reading stated 1 gr variance on these.
Or should I segregate by .3 on upper and lower keeping the .4 spread in middle as main batch? Same for short action?

 

[Ned Ludd]

I've read your posts on this subject which has me thinking strongly about weight sorting.
The part that's most dreadful is the thought of measuring volume, air bubbles n all leading to inconsistencies on my behalf.
Or is weight sorting good enough?
I did grab 10 pieces of fresh lapua brass in 30 06 and only had .9 variance.
Above post not sure who without re reading stated 1 gr variance on these.
Or should I segregate by .3 on upper and lower keeping the .4 spread in middle as main batch? Same for short action?

I should have re-iterated/clarified above that the linear relationship between case weight and case volume is not perfect...but it is a general trend. You will always find a few outliers that do not fall on the the trend line. Nonetheless, if you weigh enough cases and determine their water volume, it is not too difficult to spot the obvious places in the weight range to use as cutoffs between weight sorting groups. As I stated above, whether sorting cases by weight is actually a useful exercise depends largely on personal preference. It will always be up to the individual to decide whether their time is better spent doing something else. I have weighed and determined internal water volume for every brass prep I have done for years. I find that sorting cases by weight will typically generate more consistent internal volume than doing nothing at all.

If you wish to explore the value of this approach, I would suggest a few different things you might try initially to convince yourself one way or the other whether it is worth your time and effort. The first would be weighing AND determining water volume for a statistically significant number of cases, at least 50 to 100. Plot case weight versus case volume in a graphing program that will allow you to plot the trend line, and give you the line equation and correlation coefficient (r) for the trend line. This approach will require a fair bit of effort as accurate determination of internal water volume is a time-consuming process. That is really the reason shooter would like to use case weight, it is much easier and faster. Alternatively, you can weigh a sufficient number of cases to select some number of cases that represent the high/low extremes of the total weight range, lets say about 10 cases each. Then determine the water volume for two sets (heavy versus light). This approach is a little easier as it represents a much small total number of cases (20) that require water volume determination. Again, plot case weight versus case volume and let the program determine the trend line for the scatter plot. If there is a strong linear relationship between case weight and volume, the trend line should have a clear negative slope, and an r value greater than 0.5 or so. The closer the r value is to 1.0, the stronger the linear relationship. Technically, r should have a negative value for a line with negative slope, but not all graphing programs add in the correct sign. The reason I am making these suggestions is that accurate and precise determination of case water volume may not be simple for everyone. It requires a good analytical balance, and good technique. The less accurate and precise the water volume determination, the greater the frequency and magnitude of outliers (points not on the trend line) will be. By choosing two sets of cases at each extreme of the weight range, you should readily be able to determine whether you can readily detect a difference with your specific setup. That should help in making the decision whether sorting cases by weight is a good use of your time.

Edited to add: of course, the real reason behind sorting cases by weight or volume is to produce loaded rounds that have more consistent velocity (i.e. lower ES/SD). If you select 10 cases each (heaviest and lightest) for the exercise I mentioned above, you might as well load up all twenty cases identically, then determine muzzle velocity for all twenty. If you can't detect a statistical difference between the heaviest and lightest cases in terms of velocity, then you certainly won't be able to detect a difference between those closer together in weight/volume. Unfortunately, these are the kinds of experiments that must be done to determine whether some kind of sorting or other approach is really worth the effort. I, or others, can state all day long something works on the internet, but you really have to test and determine how something works in your hands before concluding some approach works, or does not. There are always a few people willing to make the extra effort to test and decide for themselves, but many do not.

 

[Bc'z]

I have 25 more 06 cases that need another firing to keep them in order, and be 2nd firing. On half this lot.
I've read alcohol works better than water in measuring, or should I just use water that's been left standing over nite.?
This I have to see for myself.

 

[Ned Ludd]

I use plain tap water for a number of reasons. Alcohol is not the greatest medium for measuring volume, even though some do swear by it as you mentioned. It has a greater propensity to form bubbles (which are the real killer of accurate/precise results IMO), it evaporates very quickly, it doesn't have the same surface tension as water, and it has a different density than water. Ideally, we would like to de-gas whatever liquid medium we choose to use for volume measurement, but that isn't realistic for most people. I posted a thread on the relatively simple method I use a while back; it may be of interest to you:

http://forum.accurateshooter.com/threads/case-volume-determination-pic-heavy.3896148/

Whatever method you decide on, just measure the volume for a single case, dry it out well, and repeat several times to make sure the values you get are reproducible.

 

[dstoenner]

Well in light of the fact I don't have an end mill to section a case in half, I have done some thought analysis of where the weight differences might accumulate. I agree with Ned that the primer pocket is not in consideration for 2 factors.

1) Even as manufactured they are pretty much the same or they wouldn't work

2) We all do 2 things, primer pocket unify and flash hole debur thus reducing differences even further.

So we have 3 areas that have been hypothesized as where the weight might come into play. 1) extractor groove, 2) case head thickening and 3 case wall above head to case mouth.

To make things easy I want to assume that all the cases measured come from the same lot on the same machines. I will do the analysis on a 30-06 case but the same equations apply to any other case in question.

So lets take each component factor 1 at a time and see how dimensions will effect the case weight differences. From the previous post I found that the specific density of brass was 8.53 gm per CC or 131 grains per CC. To covert cubic inches to CC we multiply cubic inches by 16.38.

1) Extractor groove

The extractor groove is a fairly complicated shape so lets simplify a little. Lets assume it is a square groove without the small ramp and that the groove height includes the point where the ramp intersects the exterior case wall. So on a 30-06 that groove would be .152 - .054 in height = .098 inches. The extractor groove is allowed to be .409 to .389 in deep. So to calculate the volume we would use the equation Pi * R squared * H But we have 2 different radii so we would calculate the outer volume and subtract the inner volume. This yields

3.14159/4 * ( .409 * .409 - .389 * .389) * .098 = .0012 cubic inches * 16.38 *131 = 2.6 gns.

But I would argue that in a case lot the extractor groove which is machined would be much less case to case variance, like maybe 1/10 of the .020 and more like .002. This still yields .269 grains.

1/10 of what we might sort on

2) Case Head thickness.

This one is pretty simple. Case head is basically .473 in diameter. If we let it be .001 inches thicker what will be the case weight contribution.

Same equation applies Pi * R * R * .001 = 3.14159/4 * .473 * .473 * .001 = .000175 cubic inches

without even going all the way to grains this factor is .01 of the extractor groove tolerance of .02 inches. This may effect to some degree but not that much.

3) Case wall.

Again to make the calculations easy lets assume a straight wall from where the case head ends to the mouth. Lets make the calculation for a .001 wall variation. Trim length on 30-06 is 2.494 and I would guess case head at .494 to make it easy so we can just get an idea. Surface area of a cylinder walls is

2 * Pi * R * H. Volume would be .001 for our wall thickness times that or

3.14159 * .473 * 2 * .001 = .002978 * 16.38 * 131 = 6.377 grains

We have to assume this is the lions share of sorting but this only represents .0487 CC's of water or .0487 gm of water.

My conclusions are that really only case wall can have a big effect on case weight. Makes sense also that the case wall is "drawn" compared to some of the other dimensions so it would have the greatest variability and hence the have an impact on volume.

Ned has made the point and this sort of shows the same thing. As weight goes up, volume goes down with small up or down uncertainty in volume because of the effects of 1 and 2 either contributing or taking away.


Now that you are all asleep I hope this was entertaining.

David

 

[Bc'z]

Not sleeping, reading and, re reading.
The math was a lil more than helping a cashier make change.

Case weight segregation makes complete sense, as I understand less volume equates to more pressure, and more volume equates to less preasure.
In my eyes and limited experience could lead to vertical stringing and or a flier.
Both of which I'm trying to eliminate.

I should be able to get remaining 25 cases fired by end of next week, putting me at 2x fired.
The test will begin.
Thanks @Ned Ludd & @dstoenner

 

[South Pender]

Your empirical findings are completely believable, Ned. Exactly what we'd expect. And your research on this topic is impeccable. Just so much better than speculation and unsubstantiated theory. So at the very least, we can use the much-easier and faster weight sorting of cases rather than having to do volume sorting. A correlation of -.75 to -.90 is easily strong enough to go with the weight-sorting.

Also, lots of good suggestions for carrying out and recording the weight-sorting operation. I think I'll do this. Whether it will improve my groups, however, is unknown at this time! I guess I'll just have to do a little research of my own.

Many thanks to you all for your thought-provoking and helpful posts.

 

[Ned Ludd]

Your empirical findings are completely believable, Ned. Exactly what we'd expect. And your research on this topic is impeccable. Just so much better than speculation and unsubstantiated theory. So at the very least, we can use the much-easier and faster weight sorting of cases rather than having to do volume sorting. A correlation of -.75 to -.90 is easily strong enough to go with the weight-sorting.

Also, lots of good suggestions for carrying out and recording the weight-sorting operation. I think I'll do this. Whether it will improve my groups, however, is unknown at this time! I guess I'll just have to do a little research of my own.

Many thanks to you all for your thought-provoking and helpful posts.

I think the most critical thing people want to know about any sorting technique is whether it will improve their scores, i.e. - is it worth the time spent? When you think about it, there are many things that different people do during the reloading process (weighing cases, weighing bullets, weighing powder to +/- one kernel, pointing bullets, weighing primers, etc.) that may be questionable investments of time for all shooters, in that it can be difficult, if not impossible, to ever prove they definitively have a beneficial effect. Nonetheless, many of us do some (or all) of these things because we believe they might help us turn out a better product in the reloading process, even if it may be difficult to quantify or prove. Whether weight sorting cases will measurably improve anyone's scores is something they have to determine for themselves. I mean, we're not typically talking about a very large total variance in internal volume within a single Lot# of quality brass.

I have always looked at some of these various techniques and asked myself the question, "How much more of my time will this technique/approach really take, in order to do it right?". As it turns out, weight-sorting cases is one of those approaches that takes very little time at all if you have a good analytical balance available, so I do it. Like anything I do, I believe there is a benefit to doing so, even if it is so small that it is difficult to prove over time. The science behind the idea is solid and the data support the theory. However, it is fair to point out that spending a lot of time addressing non-limiting sources of error may not be in everyone's best interest. I ultimately think these types of discussions are useful for helping others make informed decisions about various sorting techniques/approaches.

 

[nso123]

Just out of curiosity, has anyone ever collected data on this and other debated topics and conducted a true statistical examination? I am not talking about descriptive statistics, but something like multiple regression where you are able to assign statistical significance to a variable. It seems that there are so many possible variables that could be involved (confounders) that it would take some pretty well-considered modelling to avoid spurious results. I know the target talks back to us, but if this has not been done, I am about to fire my SPSS up and do some work in the next few weeks.

 

[Ned Ludd]

I recall one post a while ago where someone had plotted case weight versus volume for a fairly large number of cases at one time; I seem to recall it was at least 100 cases, and maybe as many as several hundred, but I don't remember the exact number. IIRC, the scatter plot was like you'd expect for a linear relationship, although not a perfect one, with a trend line having a clearly negative slope. Unfortunately, it's been some time, and I don't remember who posted the info, or the name of the thread.

The bottom line is that most people, and I include myself in that category, are not willing to determine case water volume at the level of accuracy and precision necessary for hundreds of cases at a time. Without accurate/precise water volume determination for a statistically significant number of cases, detailed statistical analysis is of limited value. So most of the data you will find online is limited to much smaller sample sets and is therefore limited in terms of the scope of statistical analysis that is practical.

 

[South Pender]

Just out of curiosity, has anyone ever collected data on this and other debated topics and conducted a true statistical examination? I am not talking about descriptive statistics, but something like multiple regression where you are able to assign statistical significance to a variable. It seems that there are so many possible variables that could be involved (confounders) that it would take some pretty well-considered modelling to avoid spurious results. I know the target talks back to us, but if this has not been done, I am about to fire my SPSS up and do some work in the next few weeks.

You don't need multiple regression analysis for the kinds of issues we're studying. Ned's reporting of a -.75 to -.90 Pearson correlation coefficient between case weight and case volume would be statistically significant at the .0005 level for as few as 15 cases measured twice (X = weight; Y = volume). With 100 cases, the p-value would be far smaller than that.

The most obvious kind of inferential statistics necessary here for comparative testing is either 2-sample t-tests or analysis of variance. For example, fire 10 or 20 5-shot groups consisting of sets of 5 rounds in which the range of case weights in each set of 5 is very small--say .25 gr. Then fire 10 or 20 5-shot groups consisting of sets of 5 rounds in which the range of case weights in each set of 5 is large--say 1.5 or 2 gr. Then compute the average group size for the 10 or 20 small-variation groups and compare it to the average group size for the 10 or 20 large-variation groups. Very simple inferential statistics--a straightforward 2-sample t-test--are all that's required. If you wanted to expand the experiment to more than two conditions--say 3 conditions with (a) small-variation, (b) medium variation, and (c) large variation, then analysis of variance will provide the inferential test. I've done similar analyses with different .22 rimfire match ammunitions to determine the best ammo in my Anschutz target rifles--comparing, for example, Eley Tenex, Lapua Midas+, and RWS R50. It doesn't take enormous numbers of groups to reach the point where the significance test has sufficient power to detect real differences. Given reasonable variability in the group sizes, I've had sufficient power with 8-10 groups of each to detect significant differences.

 

[Lannister]

I personally don't pay any attention to case weights, since there's really no way to tell where the extra weight is. First of all, are they all trimmed to the same exact length? Are the primer pockets uniformed? Are the flash holes uniformed? These all have a bearing on both the weight and accuracy potential.
The reason I don't pay any attention to total weight is it's hard to tell if the small amount of weight is in the case walls, base, rim, etc. I just uniform the brass and make sure the bullets and powder charges are uniform.
Some will disagree, but that's what I've found to work for me.
Hope this helps.

 

[nso123]

You don't need multiple regression analysis for the kinds of issues we're studying. Ned's reporting of a -.75 to -.90 Pearson correlation coefficient between case weight and case volume would be statistically significant at the .0005 level for as few as 15 cases measured twice (X = weight; Y = volume). With 100 cases, the p-value would be far smaller than that.

The most obvious kind of inferential statistics necessary here for comparative testing is either 2-sample t-tests or analysis of variance. For example, fire 10 or 20 5-shot groups consisting of sets of 5 rounds in which the range of case weights in each set of 5 is very small--say .25 gr. Then fire 10 or 20 5-shot groups consisting of sets of 5 rounds in which the range of case weights in each set of 5 is large--say 1.5 or 2 gr. Then compute the average group size for the 10 or 20 small-variation groups and compare it to the average group size for the 10 or 20 large-variation groups. Very simple inferential statistics--a straightforward 2-sample t-test--are all that's required. If you wanted to expand the experiment to more than two conditions--say 3 conditions with (a) small-variation, (b) medium variation, and (c) large variation, then analysis of variance will provide the inferential test. I've done similar analyses with different .22 rimfire match ammunitions to determine the best ammo in my Anschutz target rifles--comparing, for example, Eley Tenex, Lapua Midas+, and RWS R50. It doesn't take enormous numbers of groups to reach the point where the significance test has sufficient power to detect real differences. Given reasonable variability in the group sizes, I've had sufficient power with 8-10 groups of each to detect significant differences.

If you are getting .75 and higher on a Pearson, you run the risk of colinearity. I was not talking about just using the numbers from the case volumes, but also thinking about including other factors that might prove to be predictors. We do so much to the brass as we are getting it ready, I just wonder if something might show up in a more complex analysis. As you say, sample size does not have to be that large to get an idea of what is going on. 50-75 rounds would probably be sufficient.

 

[Lannister]

I actually used the bullets to compare several scales I have acquired over the years. From a pact electric to cheap digi and a RCBS 505 redding and a dillon which looks just like the redding but blue. all the scales seem to work well and my 1oz check weight gives all scales a thumbs up.....the pact is nice but old and takes a lot of time to zero after being off for several days and is very vibration sensitive.