Military Land Vehicle Reliability with Only 1 vehicle & Only 1 Endurance Test

  • Thread starter Reliable.Outlier
  • Start date
R

Reliable.Outlier

#1
BACKGROUND
A Brief Background for the questions: I am working as the only reliability engineer in the plant. Plant is producing military land vehicles. Each product in the plant is a military land system(vehicle) that is very complex. For an easy imagination, please look for the "Stryker IFV" and "Bradley IFV" in google images part. (note: our products are not stryker or bradley but system complexity and appearance resembles) Projects are unique for customer (generally governmental establishments like armies).

Only tool in my hand for reliability calculations is ? Only 1 Endurance Test with N kilometer with only 1 vehicle? !
My Case is: 1 military vehicle, N kilometer test until end of N?th km, quite a lot number of failures (X), some failures to be repaired, other failures to be replaced (real life :confused:)

Please help me with your answers for the questions below and Please use ?Answer for Xth Que? suffix for tracking easeness.

Thanks in advance.
Regards.

RELATED QUESTIONS FOR THE BACKGROUND ABOVE

Q1-1) my case is: I have only one vehicle, N kilometer test, quite a lot number of failures (let?s say X). So what is my quantity of samples here? Is quantity of samples in this case 1 (because of only 1 vehicle) or X (number of failures of vehicle during test until N?th kilometer)

Q1-2) Am I in a terrible statistical position because I have only 1 vehicle?s test data? What does this mean for confidence level issues; if it does?

Q2-1) each product in the plant is a military land system that is very complex so because of the high level of complexity of the system itself, may i assume that ?time to failure of the vehicle (system) is distributed exponentially at the system level?? OR ?it?s not correct to assume exponential dist. And I should analyze the endurance test data and fit a distribution for the data?? (My drawback here is: since I don?t trust 100% to our field data (failure, fault, damage etc) collecting system aka initial part of FRACAS; )

Q2-2) I don?t trust 100% to our field data (failure, fault, damage etc) collecting system aka FRACAS. If answer of Q2-1 is ?Yes, scientifically Exponential Distribution is correct & OK for a very complex system? and if I analyze the endurance test data and find that failure dist. is different quite from exponential; then what should be the results (outcome) that I should conclude?
1- Our field data is not correct?
2- Exponential Distribution assumption is not correct?
3- Our vehicle?s design is not mature yet so its failure rate is not constant so distribution is not exponential; so its failure rate is going to decrease over time because vehicle design stage is early failure stage in the bathtube curve?
4- Endurance test analysis is not correct?
5- Since some failures are repairable failures and the others are replaceable failures, there are at least 2 distributions exist for the failure data?
6- More than 1 reason (from 1,2,3,4,5)?
7- Other ?

Q3) Endurance test data must be purified (find outliers, exclude outliers) by me. Can you tell me which method(s) I MUST use for the determination of the OUTLIER data. Please clearly tell the method or methods name, because I must prove that the method selection is correct for our statistical case and my math is also correct. Please be aware that my case is: I have only one vehicle, N kilometer test, quite a lot number of failures (X)

Q4) Can you recommend some resources (statistics or reliability related books & etc) for my case? My case: 1 military vehicle, N kilometer test until end of N?th km, quite a lot number of failures (X), some failures to be repaired, other failures to be replaced
(statistics books that I?ve seen always works with lots of samples. But real world differs for me as I explained above)
 
Elsmar Forum Sponsor

Steve Prevette

Deming Disciple
Staff member
Super Moderator
#4
I do have a degree in Operations Research from the US Naval Postgraduate school that included courses for just this situation. Yes, you could make reasonable use of the data from just one vehicle, doing a Mean Time Between Failure and Mean Time to Repair analysis. Probably best to ask - do you have any contacts in the Turkish Army with working relations with the US Military? Especially the Postgraduate school (which is in Monterey, CA, USA).

A good book I have on my shelf from that curriculum is "Operational Test and Evaluation", by Roger T Stevens.

At this past year's ASQ WCQI I did meet a person currently involved in US DOD OT&E and may be able to put her in contact with you. Of course, this can get rather sensitive which is why I asked if you already have a contact in the US military.
 
A

AndrewNOConnor

#5
Some further questions for clarification:

(1) What is the purpose of the testing? Is it to meet a requirement, investigate design fixes to known failures, verification to a design standard, preparation for a customer V&V plan, estimation of the equipment life of type?

(2) What stage of the design process are you? Are these prototypes? First production run from manufacturing or quality assurance for batches delivered to customers. (The fact you mentioned lots of failures for the distance travelled suggests that your still in a design-test cycle)

To try and answer your questions:

Q1-1) If conducting this at the vehicle level, 1 sample. If at the component level then however many components on the vehicle (two headlights?).

Q1-2) Every test is left with only a confidence around what you are estimating. To know if this is an issue you need to know the test data and the confidence your company is willing to accept. If you are not familiar with the concepts of producer and consumer risk, as these will be important for a one vehicle test (particularly if the customer is accepting the vehicle based on this test).

Q1-2) It may be entirely acceptable to assume an exponential for the purposes of satisfying your test meets a standard, and if this is being delivered to a customer they are likely to accept this. In reality this is not useful and cannot be true (I think you have the same feeling which is why you?re asking the question, which many don?t). Even if we assumed that your vehicle has a useful life which came close to an exponential distribution, a test which starts with a new vehicle will have a grace period while all components are new (or burn in from manufacturing errors) until a stead state is achieved. So it again comes back to the purpose of conducting the test, for parts predictions, maintenance determinations and even calculating the ?life of type? for the vehicle, conducting a system level test and assuming exponential won?t be useful. (see further comments below)

Q2-2) You need to analyse the failures here to see what is happening. Does it have a burn in shape, or high increase in failures? Are the failures manufacturing defects? Any of the reasons you stated might be it, just need to know what stage in the design you are at and have a look at the data.

Q3) It really depends on the model you?re using. With so many failures, outliers might not affect the final conclusion. Removing outliers to me also involves an engineering assessment, and assumptions tested in doing so.

Q4) The MILSTDs and multitude of T&E handbooks will give you the exponential distribution calculations. The references you are after are those which estimate reoccurring event data or a renewal model. Note that most methods will estimate a Some books I know of are:

Meeker, William Q., and Luis A. Escobar. Statistical Methods for Reliability Data. 1st ed. Wiley-Interscience, 1998.
Nelson, Wayne B. Recurrent Events Data Analysis for Product Repairs, Disease Recurrences, and Other Applications. SIAM, 1987.
Cook, Richard J., and Jerald F. Lawless. The Statistical Analysis of Recurrent Events. 1st ed. Springer, 2007.
I tried to include some links to a Reliasoft articles, but it won't let me. Just google repairable system analysis


Other points to consider:
- If you are conducting a design-test cycle to work towards a final reliability design requirement, then you should look at reliability growth modelling. This combines the rate you discover faults with the rate you fix faults and the effectiveness of your fixes to estimate your final product reliability.

- The problem with complex equipment like this is the need to test it as long as possible into its design life. So if given the ultimatum, you should always test a single item longer (towards its expected life) than test multiple items shorter, assuming total time on test is the same. This is because shorter tests don?t uncover wear out failures which occur later in the design life. The formulas used in the MIL-STDs, or any formula which uses confidence levels on a MTBF metric (almost) always assume an Exponential distribution, which would have given you the same answer, regardless of short-many or long-one test, and in this regarded are flawed. So bottom line, you may be better off with your one vehicle of data and some engineering insight.

- There is lots more to the equation such as mission profiles, usage rates, environment, configuration consistency etc. I?ll leave these for you to consider unless you want specific help.

- A test which simply puts such a complex piece of equipment on the track and adds the failures has to be used to test integration issues and is normal for the Reliability Demonstration Test to the customer, however as the designer, if you want assurance about how reliability this equipment will be, and the risk you are exposing yourself to (not to mention parts requirement prediction and maintenance determination) then your much better off looking at this problem at the component level. This way you can determine which components will have wear out and many of the items you might already have great data for, and others may be easier to test than running a vehicle around a test track. You can also probably achieve quicker test times by focusing on the life units relevant to each failure mode. It also overcomes your problem of some components being repaired versus some being replaced.

There is a lot that can influence the answer to your question which couldn't be covered in your post, so this is just a shotgun answer of ideas. If they trigger more questions, drop me an email. I was in the Army for 15 years doing just this sort of job :D

Thanks

Andrew
 
R

Reliable.Outlier

#6
first of all, endless thanks to each of you for your answers and quick bumps. your information gave me new vision and new ways to search and learn. The reason that I am late to continue with my topic is I had a funeral & sad times. because of my lateness, I own all of you an apology.

@Mr. Steve Prevette
_________________
I don't have any contact in the Turkish Army with working relations with the US Military unfortunately. Also I don't think that there is a collaboration with US Military for these issues.

I will try to get the book you offered asap. Also if your friend needs; my email address is the reverse of text btw quotation marks: "[email protected]"
I think there is so much info that your fiend can help me

Thank you so much
Best regards

* * *

@AndrewNOConnor
________________

(1) Because of budget constraints, aims of the only endurance test are:
- demonstrate that requirements are met
- investigate immature design solutions
- reliability estimations
(2)yes, the vehicle used in test is prototype. after customer approval, serial production preparation begins. If design change requirement exists during test, design change is applied & test continues till the end of reviewed test plan. Firm has awareness of mission profile, environmental effects.Some part of requirements come from these 2 issues.
- - -
Q1-2) I know meanings of producer and consumer risk concepts. Thanks for your remark.
Q1-2) I agree with you 100% about exponential distr. ıf I was the customer, exponential distr. assumption would be the last think I will accept. Real world data doesn't promise bed of roses. Also I judge that prototype first gets more matured with some design changes or new maintenance tasks; and gets into Burn-in period so I think ı should not use exponential distr. which assumes a n useful life period for the vehicle.

What do you mean by ?life of type?? Do you mean burn-in, useful, wearout periods or smth different?
Q4) thanks for resource names. Also try to obtain these asap. There is no design-test cycle aim in our test and also there is no applied reliability growth concept unfortunately.

After reading your answer, I searched a lot about repairable system analysis. I found two models that I think (i am not sure) I can use.

1 - most suitable one > General Renewal Process (Kijima II version): 2 parameter weibull + q (rate of effectiveness of maintenance). this approach states that after maintenance action (remove&replace OR repair) state of the system is better-than-old-worse-than-new. It is very realistic I think

2 - suitable one > NHPP (Non-Homogeneous Poisson Process): If I am not wrong, this is special version of GRP (Kijima II version) where q=0 means that after maintenance action (remove&replace OR repair) state of the system is as-bad-as-old

Do you also agree that these models suit to my case? I don't know the detailed math background of these models also but first I have to decide the model. Which one do you recommend? Can you give any advice about these models if you utilized any of these?

for the General Renewal Process (Kijima II version) I couldn't clearly understand the usage of q. Does model gives a q as an output or I set my q for each failure? Second case is realistic, for example
wiper remove&replace = a repair to the vehicle = a q with a value of 0,01
engine repair = another repair to the vehicle = q with another value of 0,5
Do you know about q usage?

By the way I can't e-mail to you because forum rule says that at least five posts are required to be able to send e-mail. I have 1 post. Again my email address is the reverse of text btw quotation marks: "[email protected]". but at least I can send you pm.

Thank you so much
Best regards.

@any viewer
___________

This Conversation helps me a lot so I hope it helps you too. I thank to founders of this forum too. People should know the priceless value of being able to reach precious, added value information by the effort of helpful people that are thousands kilometers away from us. A great effort to respect.
 
Thread starter Similar threads Forum Replies Date
optomist1 Military Test Facility Certification Requirements Reliability Analysis - Predictions, Testing and Standards 5
W CE Mark for drone launcher military equipment at customer's request CE Marking (Conformité Européene) / CB Scheme 3
S Military Sampling Plan 105E vs. ISO 2859-1 AQL - Acceptable Quality Level 2
K AS9100C for Non-Aerospace Military Equipment? AS9100, IAQG 9100, Nadcap and related Aerospace Standards and Requirements 3
6 JSP 456 Volume 3 (Catering Standard for the UK Military) Information wanted Food Safety - ISO 22000, HACCP (21 CFR 120) 3
B Military / Police / EASA Regulations and Requirements EASA and JAA Aviation Standards and Requirements 4
Marc Blocked Fuel Line Botched Military Satellite Orbit Misc. Quality Assurance and Business Systems Related Topics 0
R ISO/TS 16949:2009 Applicability to Military Vehicles? IATF 16949 - Automotive Quality Systems Standard 7
K Definition Long, Wide and High - How to define for a Government (Military) Purchasing spec Definitions, Acronyms, Abbreviations and Interpretations Listed Alphabetically 3
D Military Specifications for Fasteners - Silver-Plated Self-Locking Nut - MS20500 Manufacturing and Related Processes 6
Chennaiite Is ISO/TS 16949 applicable for Defence/Military Vehicles? IATF 16949 - Automotive Quality Systems Standard 11
T Transition to AS9100 from ISO 9001 (Cost) - We are an Aerospace and Military Supplier AS9100, IAQG 9100, Nadcap and related Aerospace Standards and Requirements 31
A Transition to Military from Automotive Quality Management Various Other Specifications, Standards, and related Requirements 5
Q Using ROHS Components in Military Electronics Products RoHS, REACH, ELV, IMDS and Restricted Substances 6
C Superseded military specifications (MIL specs) Various Other Specifications, Standards, and related Requirements 19
Kales Veggie Are you familiar with EHSIR (Environmental Health Safety Impact / Canadian Military) RoHS, REACH, ELV, IMDS and Restricted Substances 1
Frank T. MIL-STD-2073-1, Standard Practice for (US) Military Packaging Various Other Specifications, Standards, and related Requirements 0
Brizilla FAA & Military Certified? Federal Aviation Administration (FAA) Standards and Requirements 3
B Electronics Die Component - Military Standard 883 vs. MIL STD 750 TM2010 B. Various Other Specifications, Standards, and related Requirements 4
C Is there a JEDEC, EIA or Military standard for Chip Scale Package Inspection Criteria Various Other Specifications, Standards, and related Requirements 5
K Military Program - Physical configuration Audit (PCA) AS9100, IAQG 9100, Nadcap and related Aerospace Standards and Requirements 8
Frank T. Mil Specs (US Military Specifications) ..... FREE in .pdf Acrobat format Various Other Specifications, Standards, and related Requirements 5
W Military Depot level maintenance or MRO forum AS9100, IAQG 9100, Nadcap and related Aerospace Standards and Requirements 12
I Some humor - Military Quotes Funny Stuff - Jokes and Humour 0
M Seeking sampling plan software and military standard design software Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
K About the Revolution, I'm an 18th century military reenactor Coffee Break and Water Cooler Discussions 6
J Seeking Military Standards Table Statistical Analysis Tools, Techniques and SPC 4
Q Military Standard 217F Questions Various Other Specifications, Standards, and related Requirements 1
C Military Personnel Support - Looking for Service Person(s) to Support Coffee Break and Water Cooler Discussions 33
C Why '105' and '414' in the military sampling plan standards (Mil-Std-)? Inspection, Prints (Drawings), Testing, Sampling and Related Topics 1
G What's your Best/Worst Military Movies? After Work and Weekend Discussion Topics 107
Claes Gefvenberg Warbirds - Military airplane fans Coffee Break and Water Cooler Discussions 83
J ISO 9001 and IDEF0/3 - Military (NATO) software maintenance/development organization Other ISO and International Standards and European Regulations 4
W Becoming ISO 9001 certified - Telecommunications and network services for military ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 4
A USA ? Military Specifications Questions Various Other Specifications, Standards, and related Requirements 2
somashekar The Land of Tests ... Book, Video, Blog and Web Site Reviews and Recommendations 2
Marc Octopus Walks on Land at Fitzgerald Marine Reserve World News 5
K Selling Land Records and Data - Quality, Legal and Other Evidence 3
S How to calculate the 'planted' area in any given stretch of land? Miscellaneous Environmental Standards and EMS Related Discussions 14
Marc Land a Million Show - Aviation Funny Stuff - Jokes and Humour 0
Wes Bucey How long before you land a new job? Career and Occupation Discussions 2
Jim Wynne Ship on land, car in water Coffee Break and Water Cooler Discussions 5
Stijloor Jaguar, Land Rover sale announcement set for March 5 or 6 World News 1
Marc How fast is the fastest you have ever driven a land vehicle? Coffee Break and Water Cooler Discussions 105
M Does "Environment" include air, water, land...etc. within the organization Miscellaneous Environmental Standards and EMS Related Discussions 4
M Percent recycle versus land fill benchmark - What percentage are you at? Miscellaneous Environmental Standards and EMS Related Discussions 1
Wes Bucey Caipirinha - New Drink Craze - And, the Land of Fruits and Nuts vs. IL Coffee Break and Water Cooler Discussions 7
S Test Specifications based on Vehicle Type Various Other Specifications, Standards, and related Requirements 0
B Fire Extinguisher Requirements for a Warehouse Vehicle Miscellaneous Environmental Standards and EMS Related Discussions 2
B List of all Auto Parts in a Commercial Vehicle Manufacturing and Related Processes 11
Similar threads


















































Top Bottom