Small Arms Reliability

  • Thread starter Thread starter .
  • Start date Start date
?

.

New member
Reliability of a mechanical device is a complex issue.

Engineers in developmental test and evalution positions view it as a series of functions termed sub-component reliablility. The operational test and evaluation communities view it as a system, or a collection of linked functions, which often include the operator.

In our previous thread there were many experiences, anecdotal reports and opinions expressed in which some incorrectly related general utility as a function of reliability.

The baseline reference point for determining reliability of a firearm is to determine a measure of failure, referred to as MTBF (mean time between failure. The threshold criteria for success and failure, plus the conditions for operation during the testing must be carefully pre-determined. Even the term failure may have sub-categories, depending on the impact the failure has on the system's ability to continue to fullfil its primary role.

All of the rifles mentioned in the previous thread are well designed and well tested firearms. However, it is unreasonable to ascribe fault to the firearm if the range of conditions change without ensuring that the appropriate changes to components, maintenance, storage, and operator protocols have been effected to ensure the required level reliability is realized.

One of the primary ingredients of success with small arms in diverse environments is the effective training and discipline of the operators. Placinge the system (firearm and operator) in a drastically different environment (changing conditions) without appropriate training in "adjustments" to the components, maintenance & preparation of the firearm, specific for the new environment is a recipe for disaster and increased reliability problems.

Whether military, LE, competitor, hunter, or recreational shooter, the same fundamental precepts apply. Use the right tool for the job, and prepare the tools and the operator for the conditions under which the job must be done!!

[This message has been edited by Mykl (edited August 01, 2000).]
 
Hi, Mykl,

Good explanation. I once had to meet with a bunch of weapons testers who had been given the task of testing software. I had a heckuva time explaining that once a software problem is fixed, it is fixed. They kept insisting that a fix would wear out, like a machinegun sear, and have to be fixed again every so many program runs.

Jim
 
Thanks Mykl, I should have known when I asked the original question that it would go like that. People get very defensive of their guns. I love my guns, but I also know that they are mechanical objects with certain specialties. And some do some things better than others.
 
Art: Thank you, kind sir!

Jim: Tell them to re-evaluate their SW metrics and system performance criteria. The only SW that changes is that which is designed to do so, as in distributive learning, neural networks and cluster recognition SW such as SWARM models.

Larry: Yup, people definitely develop irrational identity associations with "stuff" like transportation devices, environmental protection, personal ornamentation, tools, and even other people. Ain't we social critters a strange lot?
 
Minor nit. MTBF = MTTF + MTTR.

MTBF = mean time between failures
MTTF = mean time to failure
MTTR = mean time to repair

MTTF is how long a working device will continue working (on average) before failure. Then you have to repair it, which takes on average MTTR.

Thus you might really care about MTTF if your device can absolutely NOT fail - for example, it is a life-support system. Even if it only takes a minute to repair (MTTR = 1), it still means death.

Here are two contrasting examples:

Now, say my AR-15 jams on average, about once every 500 rounds (as an example). It takes about 10 seconds to clear. If I shoot one round every second, the MTTF is 500 seconds, and the MTTR is 10 seconds. So the MTBF is 510 seconds.

Now, I have a .22 pistol which had a spring break after about 5000 rounds, but it took two weeks to get a replacement part. The MTTF is 5000 seconds (assuming I shoot 1/second), and the MTTR is 2 weeks (1.2 million seconds), so the MTBF is about 1.2 million seconds.

-z
 
And an even more minor nit: you now have a single data point in determining your MTBF. Therefore your calculation is invalid. Unless you consider buying the .22 as a failure, you really don't have a MTBF. In reality you have the TTFF (time to first failure - a unique and important factor) and TTR values only.
 
Smithz,

Actually your expansion upon the the point defines Availability, not Reliability. Repair time (admin down-time + repair time = time not available for use) is not considered part of the system, but a function of support systems. The three critical measures of performance are RAM (reliability, availability, and maintainability). All of which are obviously interrelated at some level of interpretation.
 
Hey, I mentioned neither availability nor reliability in my post.

Just so you know where I'm coming from, I studied "Reliable Computer Systems" (Siewiorek & Swarz, ISBN 0137724195).

Mykl, what framework does your discussion assume?

Reliability R(t) is defined as the conditional probability that the system (device) has survived (was operational during) the interval [0,t], given that it was operational at time t=0.

Yes, my two examples illustrated availability and reliability.

Reliability is related to MTTF.

Availability is MTTF / (MTTF + MTTR).

My point was that if you really care about reliability, as defined above, you care about MTTF much more than MTBF.

Do you disagree?

-z
 
Smithz,

My "framework" work is based largely on mechanical systems, which may have a computer as a sub-component. However, the initial post is based on a previous discussion thread regarding rifles.

I think we're saying essentially the same thing, from different perspectives and for different types of systems - electronic versus mechanical.

I get the impression that you inferred that reliability was measured by MTBF. That is not the case! MTBF is a primary component (from observation data) that drives the probability calculations of reliability.

I'd agree with the your formula for defining time available, presuming MTTR is inclusive by definition of the "down-time" during which no repairs are ongoing, but the system is still unavailable. That however, would obfuscate the distinction of time lost to actual repairs and that lost due to administration to effect the repairs and return it to service.


I concur that for electronic systems, MTTF may be a more significant driver of system reliability calculations, and for a singularly critical system, such as you described (lotta folks gonna get nervous when approaching that time). However, with respect to mechanical systems, the "Between" in MTBF is useful for determining optimal service and maintenance intervals and identifying specific checks and services to be performed, which leads to the issue of maintainability...

In summary, I believe we're saying essentially the same thing about two distinctly different types of system life cycles.

Mal,
Your point is quite true, a mean is much more meaningful the greater the number of data points, and the more centralized the distribution of those points about the mean.
A single event is a point estimate, just slightly better than a SWAG. That is why mean is meaningless (pun) without the other measures of variance, distribution, and confidence interval, possibly even power of test by which the data are generated. But that ultimately drives us to the issue of statistical process control...



[This message has been edited by Mykl (edited August 04, 2000).]
 
I had a heckuva time explaining that once a software problem is fixed, it is fixed. They kept insisting that a fix would wear out

What confuses such people even further is why, after a couple years of use, we software geeks insist they have to throw out the computer and all its software even though absolutely nothing has worn out and it works exactly as well as it did when it came out of the box. I've been looking for a good way to explain that...
 
John Browning is the bane of bean counters. To get MTBF rates on his guns takes tons of ammo. To reliably determine this rate, you must fire truckloads of ammo. Now, the same rate on the M-16 takes about a case to determine. Sorry, Gene, but I think John did a much better job.
 
You must log in or register to reply here.
Back
Top