Interview Reliability Guide
Document prepared by
Indetification Number:
Delivery Date: 15/06/2010
Autor(s): Claudio Caiani Spanó
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
REVISION INDEX
REV. DESCRIPTION
REV. 0 REV. A REV. B REV. C REV. D REV. E REV. F REV. G REV. H
DATE 15/06/2010
PROJECT Internal
EXECUTED
VERIFIED
APPROVED
DATE
PROJECT
EXECUTED
VERIFIED
APPROVED
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
Index
Seção 1: Interview Data Flowchart ................................................................ 4
1.1 The Reliability Patterns ........................................................................................ 4
1.1.1 Pattern 1 – Bath Curve .................................................................................................4
1.1.2 Pattern 2 – Low Wear Out ............................................................................................4
1.1.3 Pattern 3 .......................................................................................................................4
1.1.4 Pattern 3 – Low Wear Out ............................................................................................5
1.1.5 Pattern 5 – Random Failure .........................................................................................5
1.1.6 Pattern 6 .......................................................................................................................5
1.2 Failure Pattern Selection Logic ............................................................................ 6
1.3 Questions ............................................................................................................. 7
1.3.1 Question 1 ....................................................................................................................7
1.3.2 Question 2 ....................................................................................................................8
1.3.3 Question 3 ....................................................................................................................8
1.3.4 Question 4 ....................................................................................................................8
1.3.5 Question 5 ....................................................................................................................9
1.3.6 Question 6 ....................................................................................................................9
1.3.7 Question 7 ....................................................................................................................9
1.3.8 Question 8 ....................................................................................................................9
1.3.9 Question 9 .................................................................................................................. 10
1.3.10 Question 10 ............................................................................................................ 10
1.3.11 Comments .............................................................................................................. 10
1.4 Estimating the Reliability Parameters ................................................................ 11
1.4.1 Quantity of failures in a specific period time – Exponential approach ........................ 11
1.4.2 Two lines was filled up in the table – Parameter Experimenter.................................. 11
1.4.3 More than two line filled up in the table – Free Form ................................................. 13
1.5 Estimate Logistical Time Parameters ................................................................ 16
1.6 Estimating the Maintainability Parameters ........................................................ 18
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
Seção 1: INTERVIEW DATA FLOWCHART
When performing RAM analysis and customer may not have data for all equipments or the quality
of the data is not good enough, we encourage performing Interview Reliability process, in terms
to determine the Reliability and Maintainability Distributions.
1.1 THE RELIABILITY PATTERNS
The reliability consultant must relay pattern concepts to technicians helping to characterize
failures. First analyse a failure mode qualitatively, selecting a pattern by the logic shown. Then
characterise that mode numerically, by calculating or estimating parameters governing its pattern.
1.1.1 PATTERN 1 – BATH CURVE
The pattern 1 represents equipments with more than
one failure rate behavior. The traditional shape it is
the bath curve where we have the infant mortality,
useful life and wear out. In reality we can have
equipments with more than three different behaviors during his operation life. To be able to
determine the “bath curve” with good accuracy it is better to analyze real data. For this reason
with the interview you will not be able to determine the “bath curve”.
1.1.2 PATTERN 2 – LOW WEAR OUT
Depicts age related failure resulting from deterioration
processes, but there are still considerable differences
in the behaviour of any two components subject to the
same nominal stresses. Examples are motor winding
deterioration, the choke valve on a sand producing well and erosion of pump impellers. The ‘life’
of components can be interpreted as either the recorded average or the point at which there is a
marked increase in failure probability, i.e. at the knee of the curve. Whichever is lower, the
average value or the knee of the curve, is to be used as ‘useful life’.
1.1.3 PATTERN 3
The equipment has steadily increasing failure
probability, but no single point where it can be said
the component wears out. Fatigue failures, blockage
of filters and turbine efficiency reduction through dirt accumulation are examples. The slope of
the curve is governed by the Weibull scale parameter h ranging from days to decades. It can be
steep or almost flat.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.1.4 PATTERN 3 – LOW WEAR OUT
Ooooooooo OOO
1.1.5 PATTERN 5 – RANDOM FAILURE
This means that the probability that an item will fail in
any one interval of time is the same as in any other.
Random failure is characterised by an exponential
survival distribution and constant probability of failure,
so there is no age at which scheduled rework or scheduled discard is appropriate. These items
do not have a ‘useful life’ as defined herein.
It is not possible to predict how long any one item conforming to this failure pattern will last,
however we define the MTBF as the point at which 63% of the items have failed. Items with
higher MTBFs have lower probability of failure during any given period.
Many failure modes which display random failure are not preceded by any sort of incipient failure,
or if they are, the incipient failure interval is too short to be useful, so no form of OCM is
technically feasible. Ball and roller bearings and tungsten light bulb elements display this pattern.
1.1.6 PATTERN 6
The probability of failure declines with age to reach a
constant value. It is high when the component is new
or has just been renewed, but then settles.
Compression fittings and many electronic systems,
including computers, exhibit this pattern.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.2 FAILURE PATTERN SELECTION LOGIC
Pattern 1
2%
Yes
Pattern 2
Are Failures a Do a High
Combination of No Percentage of
4%
Yes
Early Life, Use Failures Occur at
Yes Life and Wear Reasonable
Out? Consistent Age?
Are Failures
Caused by Wear No Do More Failures
Occur Shortly After Yes
Elements?
Instillation Repair or Pattern 3
Overhaul?
No Are Failures
Do Failures Increase 5%
Caused by Yes No
Steadily with Time, but
Environmental,
without a discernable
Chemical or
Sudden Increase?
Stress Reaction?
No
Pattern 4
Are Failures Mostly
Yes 2%
Random with Only a
Few Early Life
Failure?
No
Pattern 5
Do More Failures
Occur Shortly After No 14%
Instillation Repair or
Overhaul?
Yes
Pattern 6
68%
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.3 QUESTIONS
1.3.1 QUESTION 1
MOST FAILURES CAN BE CATEGORIZED INTO ONE OF A NUMBER OF DISTINCT, IDENTIFIABLE
PATTERNS. TICK THE PATTERN THAT BEST FITS THIS FAILURE MODE.
Selected Pattern Historically Comments
1 It is not
recommended to
2% determine this
pattern with the
interview.
2
4% -
3
5% -
4 It is not easy to
determine this
2% pattern with the
interview
5
14% -
6
68% -
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.3.2 QUESTION 2
To help determine the failure rate, we need to know how reliable the equipment is. Please
evaluate the failure behavior by ticking one box for each frequency.
Period Failure unlikely About evens Failure very likely Don’t know
(PF=0,01%) (PF=50%) (PF=90%)
1 month
3 months
6 months
2 years
5 years
Obs.: If you could not fill up the table with any information, try to get the number of failures for a
specific period of time.
Based in the answers you got it, filling up the table above, you can calculate the Reliability
distribution for this equipment. If for some reason you just filled up just two lines you can use the
module Parameter Experimenter on Weibull++ to calculate the parameters. If you filled up with
tree or more lines, use the Free Form in Weibull++ to calculate the parameters.
In the session XX, you can see how to proceed to perform those analyses.
1.3.3 QUESTION 3
ARE THERE ANY PRIOR WARNINGS BEFORE THE FAILURE OCCURS? TICK ALL THAT
APPLY.
Vibration Noise Leakage Performance Temperature Emission Increased Other
drop change changes lubricant (give details)
consumption
1.3.4 QUESTION 4
IF THERE ARE ANY WARNING SIGNS, CAN THEY BE MONITORED? IF SO, HOW?
Human senses Existing New instrumentation Process conditions Sample analysis
instrumentation (portable or fixed)
(portable or fixed)
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.3.5 QUESTION 5
What is the likely interval between the first sign of degradation to the failure occurring? Tick one
box per frequency.
Period Failure unlikely About evens Failure very likely
Few hours or days
1-2 weeks
1 month
2-3 months
6 months
1 year
2-5 years
1.3.6 QUESTION 6
IF THE FAILURE IS NOT EVIDENT DURING NORMAL OPERATION CAN IT BE FOUND BY
TESTING AND INSPECTION?
Yes No
1.3.7 QUESTION 7
SOME FAILURES CANNOT BE AVOIDED BY MAINTENANCE, E.G. A NAIL IS JUST AS
LIKELY TO PUNCTURE A NEW TIRE OR AN OLD ONE, SO NO OVERHAUL OR
REPLACEMENT IS EFFECTIVE AGAINST FAILURE. WOULD PLANNED REPLACEMENT OR
REFURBISHMENT MAINTENANCE PREVENT OR MITIGATE THIS FAILURE MODE?
Yes No
1.3.8 QUESTION 8
How long does it take to maintenance crew arrive to repair this equipment?
Period Failure unlikely About evens Failure very likely
Less than half hour
1 hour
2 hours
3 hours
…
…
12 hours
…
To estimate the distribution for the Time to the Crew Arrive to Repair, see the session XX.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.3.9 QUESTION 9
How long does it take to repair this equipment once maintenance has arrived?
Period Failure unlikely About evens Failure very likely
(RP = 0,01%) (RP = 50%) (RP = 90%)
Less than half hour
1 hour
2 hours
3 hours
…
…
12 hours
…
1.3.10 QUESTION 10
If spares are required, how long does it take to get the spare parts?
Period Failure unlikely About evens Failure very likely
Less than half hour
1 hour
2 hours
3 hours
…
…
12 hours
…
1.3.11 COMMENTS
DO YOU HAVE ANY COMMENTS REGARDING THE FAILURES OF THIS EQUIPMENT OR HOW
THE MAINTENANCE OF THE EQUIPMENT COULD BE IMPROVED?
Comments
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.4 ESTIMATING THE RELIABILITY PARAMETERS
Depending of the quantity of information you got it during the interview you can use
different approaches to calculate the Reliability parameters distribution. Follow the three
different approaches depending the information you got during the interview:
1. Quantity of failures in a specific period time – Exponential approach
2. Two lines was filled up in the table – Parameter Experimenter
3. More than two line filled up in the table – Free Form
1.4.1 QUANTITY OF FAILURES IN A SPECIFIC PERIOD TIME – EXPONENTIAL APPROACH
For this case, just apply the follow formula and calculate the parameter lambda (λ) for an
exponential distribution.
− λ = Number of Failures / Period of Time
If you have more than one equipment applies the follow formula:
− λ = Number of Failures / Period of Time x Number of Equipments
1.4.2 TWO LINES WAS FILLED UP IN THE TABLE – PARAMETER EXPERIMENTER
If you just filled up with two lines in the table (question x) follow the steps presented in the
example below:
Period Failure unlikely About evens Failure very likely Don’t know
(PF=0,01%) (PF=50%) (PF=90%)
1 month
3 months X
6 months X
2 years
5 years
Your data, assuming the equipment works 24 hours, 30 days per month:
• 3 months x 24h x 30 days = 2160 hours
• 6 months x 24h x 30 days = 4320 hours
Then,
Time Probability of Failure
(hours) (%)
2160 0,01
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
4320 90,00
Open the module Parameter Experimenter in Weibull++ and choose the tab “Two Unreliability
Requirements”.
Type the time and the respective probability and calculate the Weibull parameters.
In this example, based in the information you got it from the interview, the behavior of the
equipment follows a Weibull distribution with β = 7,84 e η = 3884 hours.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.4.3 MORE THAN TWO LINE FILLED UP IN THE TABLE – FREE FORM
If you filled up with three or more lines in the table, (question x), follow the steps presented in the
example below:
Period Failure unlikely About evens Failure very likely Don’t know
(PF=0,01%) (PF=50%) (PF=90%)
1 month X
3 months X
6 months X
2 years
5 years
Your data, assuming the equipment works 24 hours, 30 days per month:
• 1 month x 24h x 30 days = 720 hours
• 3 months x 24h x 30 days = 2160 hours
• 6 months x 24h x 30 days = 4320 hours
Then,
Time Probability of Failure
(hours) (%)
720 0,01
2160 50,00
4320 90,00
Open a Free Form folio in Weibull++
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
And type the time and respective probabilities”.
And calculate the Weibull parameters.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
In this example, based in the information you got it from the interview, the behavior of the
equipment follows a Weibull distribution with β = 5,83 e η = 3111,13 hours.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.5 ESTIMATE LOGISTICAL TIME PARAMETERS
To estimate the probability distribution for either Time to Crew Arrive to Repair or Time to Get the
Spare Part, follow the example presented next.
Period Failure unlikely About evens Failure very likely
2 hours x
6 hours x
12 hours x
Then,
Time Probability
(hours) (%)
2 0,01
6 50,00
12 90,00
Open a Free Form folio in Weibull++, type the data and choose the Weibull distribution.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
And calculate the Weibull parameters.
In this example, based in the information you got it from the interview, the probability follows a
Weibull distribution with β = 6,32 e η = 8,31 hours.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
1.6 ESTIMATING THE MAINTAINABILITY PARAMETERS
To determine the maintainability distribution follow the example presented next.
How long does it take to repair this equipment once maintenance have arrived?
Period Failure unlikely About evens Failure very likely
(RP = 0,01%) (RP = 50%) (RP = 90%)
Less than half hour
1 hour x
2 hours x
3 hours x
Then,
Time Probability to Repair
(hours) (%)
1 0,01
2 50,00
3 90,00
Open a Free Form folio in Weibull++, type the data and choose the Lognormal distribution.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]
And calculate the Lognormal parameters.
In this example, based in the information you got it from the interview, the maintainability follows a
Lognormal distribution with μ = 0,7697 e σ = 0,2122.
Rua Carneiro da Cunha, 303 12º andar - São Paulo – SP – Brazil
Telefone: +55 11 21775456 FAX: +55 11 21775452
[Link]