1,800 per month @  USD 71 ea. x 12 months=

Source of the 1,800 number.

I believe the OP was ALL AR-type bolts, not just military (at least that's how I read it). In the subpopulation of Army weapons, you may very well see something close to a normal distribution, as the number of weapons actually deployed and actively used in combat is likely very small compared to the total inventory, and would not influence the total distribution too much; and all are (theoretically) subject to the same ammunition, maintenance and training procedures. When you add in the other subpopulations, though- civilian, other service branches, police- you have distinct groups encountering different sets of stressors. If different sub-populations have different stress factors, you won't see a normal distribution for the entire population. Within each sub-population, assuming every bolt experienced the same mix of stressors, you would indeed see a normal distribution. However, because the separate subunits may have unique factors that affect bolt life, the expected distribution of the entire population would be multimodal, provided the sub-populations are of a statistically significant size. It probably wouldn't be clean enough to see separate peaks, but it would likely distort the distribution curve beyond anything that could be called normal. If it did come up with a clean normal distribution it would be very interesting, as it would imply that differing deployment conditions, ammunition, and maintenance had no significant effect on bolt life.

Of course, this is all rather futile, as in order to come up with data for any sort of distribution, a statistically significant sample of bolts (likely thousands, certainly hundreds at least) would have to be tested to failure under each of a large number of sets of controlled circumstances.

Actually, I think I have a better idea of what the OP is talking about than anybody.

It was based on the Army data someone provided that stated the average bolt life was 12,000 rounds for the Army.

(Besides, Army sort of keeps track of bolt life, and can estimate how long they last.  How would you even start to gather that data for the civilian population of bolts.)

Very interesting topic, I love technical discussions like this.  

But have to ask the probably stupid question, what about the carriers?  Under what circumstances would that be replaced?  Speaking strictly within .mil

There is a problem with that 1,800 bolts per month consumption rate.

The number of bolts, breech, P/N 8448510, (stripped bolt) delivered in the last ten years was 255, at a total contract price of $13,333.95 ($59.26 each).

The number of Bolt Assembly, P/N 13004787, (complete bolt) delivered in the last ten years was 50,000, at a total contract price of $2,209000 ($44.18 each).

If they were going through bolts at a rate of 1,800 a month, the last ten years of bolt deliveries would have been consumed in just over two years.

For combat arms units.  Replace bolt carrier at 60K, ref SOFWEP-07-G12P-00032-00 Rev 1
SW370-CC-TRS-010 Rev 1
TECHNICAL REPAIR STANDARD
FOR
CARBINE, 5.56 MM, M4A1

Appendix F, Table F-1

COMPONENT  PN 10K 15K 30K 45K 60K 75K 90K 105K
Upper Receiver and
Barrel Assembly
12997148 I I I I I I R
Bolt Carrier Assembly 13004788 I I I R I I
Pin, Firing 8448503 I I I I I I I I
Pin, Firing, Pin
Retainer
8448504 I I I I I I I I
Pin, Grooved, Headed 8448502 I I I I I I I I
Carrier, Assembly, Key
and Bolt
8448505 I I I I I I I I

Notes are I is Inspect, R is replace


CD

General wear, stripped threads, over-size bore, plating deterioration, cracking, the usual stuff.

Generally carriers don't wear out as quickly as other parts.

Actually most low cycle fatigue failure is skewed to the left more often than not.  If your stressing the part hard enough that failure occurs relatively quickly than exploitation of flaws will skew this left.  Go get better data this is not an assumption you can make and still have a good discussion.

OK.  From the some Army reliability testing ("Baseline Reliability for the M4 Carbine, M16 Rifle, and M249 Machine Gun"):

For new manufactured M4 Carbines, the 10% failure is 7000 rounds, the mean part life is 11,300 rounds, and the 90% failure is 17,000 rounds.

For new manufactured M16 Rifles, the 10% failure is 11,000 rounds, the mean part life is 15,000 rounds, and the 90% failure is 19,000 rounds.

That's a pretty normal distribution.

The only time it gets skewed to the left (in this case) is when you get remanufacturing into the mix, as some parts that still have life are reused, but are nearer the end of their life, so they fail soon after the remanufacture.

Let me say that I do appreciate the detailed explanation

I just wanted to offer a perspective on your Pop Quiz.



Don't care.

Before I bought an AR a couple years ago, I had a Mini-14 that I had bought shortly after I was old enough to legally buy it and it has been with me for nearly 40 years.  In that time, it, and another Mini-14 I subsequently bought for my father, have gone through less than 5,000 rounds, total.  I figure the bolt life of my ARs will be of concern only to my heirs - and I the same can be said of every other AR owner I know.

Join date, post count. And round count.

Outstanding info, thanks for sharing

Question -

Bolt Carrier Assembly, P/N 13004788 is the entire bolt carrier assy to include the bolt, firing pin, retaining pin, cam pin, and bolt carrier (ie, all the other parts noted in this table).

If you replace the entire assembly at 45,000 rounds, why do all the other sub-components of that assembly say "inspect"?  By replacing the higher assembly, all the sub-components would get replaced as well.

If it 60,000 rounds, you missing an  "I", 60K is at the fifth place, the "R" is at the forth place or 45K.

It's a formatting thing. It's 60,000 for the bolt carrier assembly.

Hundreds of thousands of bucks is absolutely nothing to the Feds.

Single agencies may misplace - as in lose entirely - billions in a single year.

OP, what is your justification for assuming the standard deviation is 1300 rounds?

does the strength of the ejector spring have any effect on how much chambering force is transmitted to the bolt face and lugs?  I recently changed a worn out ejector spring on my AUG.  Previously it had been really easy to reinsert and get the roll pin in to hold the ejector.  But when I put the new ejector spring, it was a fucking beast to push in and hold to the get the roll pin in place.  I also had been able to slightly ride the charging handle when chambering a round with no problem. With the new spring, I have to let it fly full force to get the round to chamber. 

On an AR, does a fresh ejector spring provide any noticeable resistance when a round is being chambered?  If so, as the spring weakens (mil-spec is 3,000 rds) is there more force on the bolt face and lugs and could that cause sheared lugs or cracks?


(old spring- left; new spring- right)

It a reasonable round number that fits.  (See the post with the 10% and 90% failure counts.)

Although, the point of the original post was to show that the "average life of 12,000 rounds" and the often quoted "5,000 round life" are not incompatible, depending on how you want to deal with possible bolt failures.

I also noticed that all of the component parts of the Bolt Carrier Assembly, P/N 13004788, are listed in that table EXCEPT the Bolt, Breech Assembly, P/N 13004787.

What is the bolt's inspection/replacement schedule?

Bolt is a different category and listed at 10K / R.

That's what I thought. And honestly why I'm having a hard time with the analysis.

First, a Weibull curve is much more appropriate, since it more easily models events vs. time.  a Normal distribution is better suited to events in a population.  Yes, one can define "events" as "failures at X rounds", but that in itself is a derived number.

A different way to look at it is that, per your fit, while a tiny number of bolts fail at <5000 rounds, some do.  And per the curve, that means a VERY tiny number of bolts will fail at <0 rounds. That is an impossibility in this context - we are looking at failures vs. round count where the lowest possible value is 1. And a model, while it doesn't have to be perfect, CANNOT allow for the impossible - if it does, the model is inappropriate.

My second issue is that mean and standard deviation and other parameters are supposed to be derived from the data itself.  From the looks of it, there are 2 pieces of actual data - average round count at bold failure=12000, and another less well attributed number of 5000 round life.  So if we *assume* that 5000 rounds is .01% failure rate, you can back calculate the rest of the curve.  But you say that number is the "safe fatigue life."  That isn't a defined number - it is different per application, industry, code, etc.  So if 5000 is actually 99.9% (the "Six Sigma" number if I recall correctly), that will change the SD and the number at the upper end.

I appreciate the thought and effort that went into this.  And I'd really love to see the dataset used to get the 12000, etc. But the curve and numbers you've posted just seem too loose.

If you can find the report on the baseline reliability of the M4 carbine, that's where the 12,000 round average bolt life comes from.  For M16s its 15,000 rounds.  That has all the methodology information.

Is that 12k with 855a1? What's it with the lower pressure stuff?

No, what he's describing is the mean (which is the most common form of average).

The median is the middle number out of the set. The median of 1, 2, 3, 4, 1000 is 3.

The mean is the sum of the numbers divided by how many there are. The mean of 1, 2, 3, 4, 1000 is 202. Fatigue cracking isn't a rapid event, it's generally something you can spot well in advance of catastrophic failure.



It grows slowly (generally polishing itself to some extent) for a long time, and then eventually fails rapidly (leaving a rough fracture). Is that just for bolts or are there other failures in there too?

I'm a little surprised at how much shorter bolt life is for M4s, I guess the unlocking adds significantly to the stresses on the bolt.

If it's just for bolts, I'd be surprised. I would expect an M16 bolt to last significantly longer than an M4 bolt. At least more than the claimed 2k round difference.

For anyone wanting to gamble on better materials, SRC has their S7 DLC bolts for $45.

http://store.srcarms.com/product-p/xpb.htm