Notice: This technical discussion forum has been established by the SRE to assist all reliability engineers, not just SRE members. To post your technical problem, solution, question, or answer here, send an e-mail to SRE webmaster. Your e-mail should follow the format of those already posted. Postings are accepted from anyone, as long as they relate to the reliability field. It works best if your posting contains contact information for possible follow-up. But, if necessary, postings can be anonymous - just so state in your email! Also, let us know when you do get workable solution or answer. We want to post it so everybody will benefit.

Problem/Question (December 12, 2001):
Hi,
I am looking for a formula to calculate a 1-out-of-3 System-MTBF. It shall be based on MIL-HDBK-217 / Markov.
Does anybody know?

Guenther Ebner
E-mail: ebner@siemens.com>

Solution/Answer: Can you help?

Problem/Question(January 20, 2006):

Answer (January 20, 2006)

RELEX's reliability prediction software has a Reliability Block Diagram (RBD) capability with Markov Simulation based on MIL-HDBK-217F, you might contact them to send you a DEMO, then you can decide whether or not it is applicable to your need/s. Good Luck.

Hank Sofian

Problem/Question (November 22, 2001):
I'm looking for a good general reference book for someone relatively new to reliability engineering. I have decent background in statistics so I'm not really looking for a complete beginners book. Likewise, I'm not looking for a text where the reader would need to have a Ph.d. in applied statistics to understand its contents. Just a good general text that covers the "applied" portion of reliability engineering.

Any help would be greatly appreciated.

Thanks!

David Winter
E-mail: davidwinter@prodigy.net>

Solution/Answer: Can you help?

Problem/Question (October 1, 2001):
Hello,

Please, I need some information about Reliability Techniques (Spanish books, articles, papers, magazines, etc.). Do you know if in Venezuela there are people who has had experience about this?. If So, please let me know how I can contact these people. My Post Grade Thesis is " Reliability Technique Development in Maintenance Projects". I wonder if you can help me whit this Project.

I am an Electrical Engineer and I work in the Maintenance Department in a Fish and Food Processing Factory (Moves Mariguitar S.A.)

Marcos Millán
Ing. Electricista
Departamento de Ingeniería y Mantenimiento
Alimentos Margarita C.A.
Playa el Mamey,
Mariguitar, Edo. Sucre
Phone: + 58 + 293 + 8391224 - 8391222 - 8391054
Fax: + 58 + 293 + 8391174
E-mail: marcos_millan@mavesa.com>

Solution/Answer: Can you help?

The discussion thus far:

(December 11, 2001):
Here are two articles in Spanish.
"Raising the Field Reliability Benchmark," by L. L. George and Eva Langfeldt, ASQ Reliability Review, Vol. 20, No. 3, Sept. 2000, translated and reprinted in Gestion de Activos Industriales, Issue No. 14, Nov.-Dec. 2000.
"MTBF vs. Age-Specific Reliability Prediction," by L. L. George and Eva Langfeldt, ASQ Reliability Review, Vol. 21, No. 2, June 2001, pp 13-15. Translated and reprinted in Gestion de Activos Industriales sometime after Sept., 2001.

Contact Name: Larry George, PhD
Fellow and CRE American Society for Quality
Phone: 925-447-4969
Email: pstlarry@attbi.com

Problem/Question (September 18, 2001):
I am working on a project with the Army to try to increase the reliability of their equipment over a short period of time. I am currently trying to compare current Army reliability parameters and requirements to similar commercial equipment . Is there any place, web page, book, or paper where I can find current reliability parameters and goals for heavy equipment, diesel engines, generators, or heavy trucks without compromising proprietary information? Generic standards or standards used in studies would be sufficient for my needs.

Don Sapp
Plexus Scientific Corporation
1900 N. Beauregard St
Suite 300
Alexandria, Va 22311
Phone: 703-845-8492, ext 116
Fax: 703-845-8499
E-mail: dsapp@plexsci.com

Solution/Answer: Can you help?

Problem/Question (July 13, 2001):
Whilst performing fault tree analysis with the relex package I have encountered these queries. Any readers that may have met and overcome them, your help/guidance would be appreciated.

1. During fault tree analysis changing individual components from repaiable to non-repairable and vice versa, does not always provide consistant total system reliability results.

2. When printing reports the probability box overlaying the logic gate symbols disappears and cannot be printed.

Any help on either of these two queries would be gretaly appreciated.

Engineering
E-mail: Contact@msi-eng.freeserve.co.uk

Solution/Answer: Can you help?

Problem/Question (July 11, 2001):
What product is most popular in the industry for Failure Reporting, Analysis, and Corrective Action System (FRACAS)? I am doing trade study and comparing different solutions for our FRACAS. We can start from scratch by Oracle or we can buy some off-the-shell products like V-FRACAS from a company in Israel, Lambda from England, and TrackWise from Sparta Systems Inc in NJ.

Thanks in advance.

Xueli Wang
Medrad Inc
E-mail: xwang@medrad.com

Solution/Answer: Can you help?

Problem/Question (June 19, 2001):
Hi,

I am working for a electronics company and I am interested in a procedure to calculate the life fit rate data for one of our new products.

Regards,

Sorin Weissman
Melcor Corporation
1040 Spruce Street
Trenton NJ 08648
Phone: 609-393-4178
Fax: 609-393-9461
E-mail: sweissman@melcor.com

Solution/Answer: Can you help?

The discussion thus far:

(July 20, 2001):
Here are some alternatives for estimating failure rates, in order of my preferences if I was your customer:
1. MIL-HDBK-217 cost = $0, labor = depends on bom
2. BellCore TR332, http://www.telcordia.com, cost ~ $1000, labor = same as MIL-HDBK-217
3. RELEX, http://www.relexsoftware.com, cost ~ $4000, labor = less except for learning to use the software
4. PRISM, http://rac.iitri.org, cost = ?, labor ~ same as RELEX
5. Rob Poltz, http://www.designanalytx.com, cost = consulting fee, labor = 0
6. Would you like a prediction of the age-specific reliability and failure rates? For example:
P[Life > 0] = 1-P[DoA]
P[Life > k quarters], k = 1, 2,...to useful life
P[Life > warranty]
P[Life > useful life].
If so, let me know, and I'll send you the article on how to do it. Cost = free for samples. Labor = you have to provide field data for comparable products.

Contact Name: Larry George, PhD
Fellow and CRE American Society for Quality
1573 Roselli Drive
Livermore, CA 94550-5852 USA
Phone: 01 925 447 4969 v and f (Please call first to fax. Thanks)
Email: pstlarry@home.com

Problem/Question (May 31, 2001):
Hello,

I am looking for an equation to convert Field Failure % (AFR) into Mean Time Between Failure (MTBF). Any help would be greatly appreciated.

Thanks,

Andrea Mourning
Quality Assurance Engineer
E-mail: amourning@opuswave.com

Solution/Answer: Can you help?

The discussion thus far:

(July 20, 2001):
Sigh. You can't convert AFR into MTBF, but people do. The formula is:
MTBF = 1/AFR
The units are years. This formula is valid only if the field failure rate is constant. Otherwise, the relation between AFR and MTBF is tenuous. Field failure rates aren't constant. Sorry.

For example, a formerly major computer company tested new products and estimated MTBF from the results. Their 1/AFR was actually in the same order of magnitude as their reported MTBF. Due to pressure they were lucky to get a small sample and only a few weeks to test. For various reasons, a lot of failures occurred, most of which were not counted. When pressed, the company admitted that the reported MTBF was actually the lower 5% confidence limit on the MTBF.

If you would like to know the real field reliability and failure rate functions of your products, send the periodic production and returns data underlying your AFR calculations, and I will send back nonparametric estimates of age-specific field reliability and failure rate functions, free. Those estimates are actionable. That means you can do something about reliability. AFR charts are for wallpapering cafeterias and halls.

Contact Name: Larry George, PhD
Fellow and CRE American Society for Quality
1573 Roselli Drive
Livermore, CA 94550-5852 USA
Phone: 01 925 447 4969 v and f (Please call first to fax. Thanks)
Email: pstlarry@home.com

Problem/Question (April 24, 2001):
Can somebody please point me towards a spares analysis procedure. I need to suggest the numnber of spares required for a mission, based on item MTBF and the mission duration.

Thanks in advance,

Richard Smith
E-mail: Smith.Richard@littonlsl.com

Solution/Answer: Can you help?

The discussion thus far:

(May 8, 2001):
To determine number of spares at the desired confidence level I would use the assumption that if the times to occurrence of an event are exponentially distributed, having rate lambda, then the number of events in any length t is Poisson distributed with mean lambda*t. Solving for the Poisson distribution we can arrive at this identity F(k) = P(k or fewer failures). These results can be used to determine the prob. of occurrence of a specific number of failures during a mission, or to calculate the number of spares required when units have constant and identical failure rates.

Contact Name: Sekhar Vaidyanath
RAMS Engineer
US&S
Phone: 412-688-3065
Email: svaidyanath@switch.com

(June 25, 2001):
Have you looked at using software. In particular, APT Spares. (Asset Performance Tools)

It determines the amount of spares that should be kept considering the financial risks associated with that.

If this is maybe an option, let me know and I'll inform you about the publishers of this product.

Regards.

Contact Name: Wayne Erasmus
Email: wayne.erasmus@sasol.com

Problem/Question (April 20, 2001):
I am a RAMS Engineer working for a systems company and my question is as follows. What is the failure rate of a Hall Effect Switch?

Sekhar Vaidyanath
RAMS Engineer
US&S
E-mail: svaidyanath@switch.com
Phone: 412-688-3065

Solution/Answer: Can you help?

The discussion thus far:

(May 7, 2001):
The following table shows nonparametric, maximum likelihood estimators of the monthly field return rate of Hall-effect crankshaft or camshaft sensors used in automotive applications.

Not all returns were failures, but I suspect most were, for this sensor. The monthly return rate is not constant; it is P[Return next month | Survive to specified age]. This Hall-effect sensor suffers from premature wearout, presumably caused by design defect. This sensor was the worse than non-Hall-effect sensors.

I presume a Hall-effect switch consists of a Hall-effect sensor and a switch. To estimate failure rate of Hall-effect switches, add the monthly failure rate of the switch part to the failure rates in the table.

I hope your Hall-effect switches are better than these sensors. If you would like to find out, send ships and returns data, and I will send back nonparametric, maximum likelihood estimates of reliability and failure rate functions, free.

Contact Name: Larry George
PhD CRE (Certified Reliability Engineer)
1573 Roselli Drive
Livermore, CA 94550-5852 USA
Phone: 01 925 447 4969 v and f (Please call first to fax. Thanks)
Email: pstlarry@home.com

Problem/Question (April 17, 2001):
I am working with the Corporate Quality Office of an automobile company. I am trying to gain insight in the field of Reliability Engineering. For some time now, I have been trying to understand Weibull analysis and have tried to apply it in our company to get a 'hands-on' feel. By doing so, I am now comfortable with the methodology. But, at the same time there are a few questions raised:

1. When the scatter of the readings is less, and the beta value obtained from the plot is high (>1), what is wrong with it? What does the wearout phase really mean?

2. In any sample, where there are consistent readings (less scatter), the Weibull plot will always be a near vertical straight line.(I,e beta much more than 1). Is it bad or good?

3. What should a designer do, after getting beta>1 in a Weibull plot?

4. For some reasons, if the testing has to be stopped in between, will it do? Can we just start counting the cycles after resuming the test?

Can you help or guide me?

Shashank Kotwal
Phone: 091-22-887 4601, Ext. 2768
Fax: 091-22-887 1695
E-mail: kotwal.shashank@mahindra.co.in

Company Name and Address:
Mahindra & Mahindra Ltd.
Corporate Quality Office (CQO)
Automotive Sector
Akurli Road, Kandivili (E)
Mumbai-400 101
India

Solution/Answer: Can you help?

The discussion thus far:

(May 16, 2001):
Since I havent done any analysis with your data, I cant answer your specific Weibull questions. In general, Weibull of Beta = 1 can explain exponential type failures and B>1 can explain wear out type failures. For example if you study a large population you will find early failures, late failures and chance (no specific pattern) failures and a mixed Weibull analysis of this data set can separate the 3 sub-populations to get a "bath-tub" curve.

To answer your questions:
1. Mathematically, I don't see anything wrong in getting large Beta values. Steep beta's (i.e., large values for beta) have small variations in the data (and thus the Cv is small). Shallow beta's (i.e., small values for beta) have large variations in the data (and thus the Cv is large). For most issues, you want steep betas with small Cv. Cv is the coeff. of variation.
2. I don't have an answer to your good or bad question. If you have worked everything out correctly it is good.
3. I wouldn't worry about beta > 1 as long as it meets the reliability goals.

Contact Name: Sekhar Vaidyanath
Systems RAMS Engineer
US&S
Phone: 412-688-3065
Email: svaidyanath@switch.com

Problem/Question (April 16, 2001):
Hi. I'm trying to find a formula to convert thermal stress, I need to submit a sensor at -40 deg C for 25 hours, my equipment only gets down to -35 degs C, how long do I need to submit mi sensor to this temperature so the stress keeps the same? Thanks for your support.

Pedro Corral
E-mail: pedro.corral@delphiauto.com
Solution/Answer: Can you help?

Problem/Question (April 11, 2001):
I am a engineer in the reliability dept of a nuclear generating facility. I am looking for a reliable source for metalurgical failure analysis of anti friction bearings. Anyone know of any?

Doug Long
E-mail: douglas.long@wepco.com

Solution/Answer: Can you help?

Problem/Question (April 10, 2001):
I am looking for an equipment which can generate negative voltage pulse of -300 V up to -1000V for testing the automotive products. Does anybody know which company selling this type of equipment ?

Many thanks

CT Ewe
Penang, Malaysia
E-mail: chintean.ewe@my.bosch.com

Solution/Answer: Can you help?

The discussion thus far:

(November 8, 2001):
There are a number of manufacturers of surge or pulse generators. I would recommend you contact

Problem/Question (April 9, 2001):
Hi, I am a quality engineer and I have a question geared towards Maintainability. I just came across a requirement to achieve a 90th Percentile Maximum Corrective Maintenance Time (MAXCMT) of not greater than one hour at the Operational Maintenance Level. Could anyone explain to me what this actually means.

Jeff Yuan
Phone: 905-792-2115, x417
E-mail: jyuan@atlantissi.com

Solution/Answer: Can you help?

Problem/Question (April 2, 2001):
I am an electrical engineer who has spent the last four years doing reliability. I want to understand, quanitatively, the reliability impact of circuit board modifications after assembly.

Instinctively, I have to believe there is an impact, but I can find no quantitative information anywhere. Modifications include track cuts, wire bonds, lifting component leads, and adding components to the assembled board.

Does anyone know of a study where FITs have been estimated for circuit pack modifications?

Mary Jo Skeet
Phone: 403-769-4804
E-mail: mjskeet@nortelnetworks.com

Solution/Answer: Can you help?

Problem/Question (March 29, 2001):
Goodday to you,

My son is a first year Electrical (High Currency) student. He is writing a Theory Paper and would appreciate assistance indefining the term - Fault Current - Could anybody help? Information urgently required.

Many thanks

Celia Horn
GFS Electrical
P O Box 2389
Pongola
Kwa-Zulu Natal
3170
South Africa
E-mail: pongotoy@pga.dorea.co.za

Solution/Answer: Can you help?

Problem/Question (March 16, 2001):
Hi, we are involved in the development of OLEDs. I´m interested in accelerated reliability testing for displays. Can anybody tell me which standards are used here?

Thanks!

Christine Kallmayer
Fraunhofer IZM
Gustav-Meyer-Allee 25
13355 Berlin, Germany
E-mail: kallmayr@izm.fhg.de

Solution/Answer: Can you help?

Problem/Question (March 1, 2001):
I am an electrical engineer preparing a series of accelerated wear tests using extreme temperature cycling methods. Can anyone tell how to correlate the temperature cycling time to real life wear time?

Sincerely,

Kevin Bodily
Electrical Engineer
Varatouch Technology
Phone: 916-331-6300, x322
Fax: 916-338-8255
E-mail: kbodily@varatouch.com

Solution/Answer: Can you help?

Problem/Question (February 11, 2001):
I am a student of electrical engineering, Gadjah Mada University Yogyakarta, I'm a beginner in reliability. I am working on a project to evaluate reliability of power distribution system on an hospital. I want to know:

1. How to get the failure rate of each component of the system and the hospital don't have any record about the failure of each component?
2. Is there any standard of failure rate of the component?

Could somebody help me to provide this information or any internet address to get this information? Thank you very much and best regards.

Milladunna D. Dzikron
Gadjah Mada University
Yogyakarta, Indonesia
Phone: 62-274-902202
E-mail: midun96@lycos.com

Solution/Answer: Can you help?

The discussion thus far:

(March 16, 2001):
1. Generally accepted accounting principles require ships (installations, production, purchases, etc.) and returns (failures, complaints, repairs, spares sales, etc.) counts. Ships and returns counts are sufficient to make nonparametric estimates of age specific field reliability and failure rate functions. In other words, accountants have sufficient data! Please see http://members.home.net/pstlarry/Table.htm for a sample data entry form. Accountants record ships as how many units were placed in service, because they paid for them. They record returns because somebody had to pay for the repair or replacement costs. Counts are grouped by accounting intervals, usually calendar quarters.

2. There are several sources of failure rate estimates. They assume failure rates are constant, independent of age. Failure rates are rarely constant, which makes life interesting and provides actionable information to reduce failure rates if economically worthwhile.

If you can find ships and returns counts (It's not always easy.), then send them, and I will send back nonparametric estimates of age-specific field reliability, free.

Contact Name: Larry George
Email: pstlarry@home.com

Problem/Question (January 30, 2001):
I am working on a project to comapre how Department of Defense (DOD) apply Relaibility Availability Maintainability Engineering Management as compared to Private Industry. I am searching for any publication or any paper has been written in the past on this topic. Could somebody help me to provide this information or any internet address to get this information?

Thank you very much and best regards.

Syed I. Ahmad
US. Army AMCOM
Huntsville, Al 35898
Phone: 256-876-2646
E-mail: Syed.Ahmad@rdec.redstone.army.mil

Solution/Answer: Can you help?

Problem/Question (January 29, 2001):
Hi, I am using the MIL-HDBK-217F N2 for calculating Hybrid failure rate. however, using MIL-HDBK-217E gives a much lower failure rate. For example take one device, GaAs FET device in a microwave application. The Failure rate calculated using 217F N2 gives a value which is of order of 24 times higher than value calculated using 217E. One reason is, the fact that in the 217F model, the PiQ and PiE factors are not applicable when calculating the device failure rate, where as they are accounted in the 217E model.

Can anyone give field failure rate data or experience (field failure rate data) to show that the 217E Hybrid model is more realistic and gives values comparable to field failure rate data on hybrids. Since the contractual document is MIL-HDBK-217FN2, I am not allowed to use MIL-HDBK-217E, which gives me a more favourable result for Hybrids. Sincerely,

B.Rao
E-mail: Rao.B@ems-t.ca

Solution/Answer: Can you help?