1. Relex Reliability Software “the intuitive solution
Relex Reliability Software “the intuitive solution!” Relex Software Corporation 1

2. What is Relex? A Powerful Reliability Software Tool…
performs efficient reliability analysis
uses multiple analysis techniques
provides advanced features

3. Relex Is Uniquely Qualified
Reliability Engineering Experience
Commercial
Military
Software Development Experience

4. Relex Reliability Software “the intuitive solution
Relex Reliability Software “the intuitive solution!” Relex Software Corporation

5. Introduction to Reliability Prediction

6. Reliability Predictions
What is a Reliability Prediction?
Calculation of failure rate (MTBF)
How is it Calculated?
Based on established reliability model

7. Sample Relex Reliability Prediction calculation results
Reliability Measures
Failure Rate ()
Mean Time Between Failures (MTBF)
Reliability
Availability
Sample Relex Reliability Prediction calculation results

8. Failure Rate Defined As: Units: Rate of Occurrence of Failures
Number of Failure in Specified
Time Period
Units:
Failures per Million Hours
Failures per Billion Hours (FIT Rate)

9. MTBF Defined As: Units: Mean Time Between Failures
Number of Hours to Pass
Before a Failure Occurs
Inverse of Failure Rate*
Units:
Typically expressed in Hours
*Constant Failure Rate Systems

10. Reliability Defined As: Units:
The probability that an item will perform a
required function without failure under
stated conditions for a stated period of
time
Units:
Probability Value (0-1)

11. Availability Defined As: Units: The probability that an item is in an
operable state at any time
Units:
Probability Value (0-1)

12. Reliability “Summary”
Failure Rate -- number of failures in time
MTBF -- average time between failures
Reliability -- takes into account mission time
Availability -- accounts for repairs (MTTR)
and downtime

13. The Bathtub Curve and Reliability

14. The Bathtub Curve
Represents failure rate tendencies for the lifespan of an item
Failure rate varies in different phases of life

15. Three Phases of Life Infant Mortality Region Wear-Out Region
Constant Failure Rate Region

16. Bathtub Curve Graph of Failure Rate vs. Time
Considers three phases of life
Represents lifespan of item
(i.e. 15 years for a car)

17. Bathtub Curve –Illustration–
Infant Mortality
Wear Out
Constant Failure Rate
Failure Rate
Time 17

18. Reliability Models

19. Influences to reliability / Model-parameters
Production maturity
Design &
construction
Storage conditions
Production factors
Transport conditions
Material- selection
Electronic component
Application- temperature
Operating
conditions
Application factors
electrical stress
mechanical stress
Climatic environment

20. Relex Prediction Models
MIL-HDBK-217 (FN1, FN2 )
Telcordia (Telcordia 1, Bellcore 4,5,6)
Prism: RAC model (Process Grades, Bayesian)
NSWC-98/LE1: mechanical model
HRD5: British telecomm model
CNET 93: French telecomm model
299B: Chinese standard
Relex allows the user to use multiple models within one project and use functionality across models (i.e. use Prism process grade factors on 217 predicted failure rates, use Bellcore methods on 217 calculations, etc.)

21. MIL-HDBK-217 Original standard for reliability
Reliability math models electronic devices
Used commercially & in the defense industry
Currently at Revision F Notice 2

22. Parts Count A section of MIL-HDBK-217
Provides simpler reliability math
Typical Uses:
Used early in the design process
Used to acquire a rough estimate of reliability

23. Telcordia (Bellcore) Originally developed at AT&T Bell Labs
“Modified” MIL-HDBK-217 equations
New equations represented what their equipment was experiencing in the field

24. Telcordia (Bellcore) (cont.)
New model with new feature
Account for “real data”
Burn-in, Field, Laboratory testing data
Popular standard for commercial companies

25. Mechanical
Based on the Handbook of Reliability Prediction Procedures for Mechanical Equipment, NSWC-98/LE1
Provides models for various types of mechanical devices including springs, bearings, seals, etc.
New and unique standard

26. CNET & HRD5 Used in Europe Reliability models for telecommunications
Current Versions:
HRD - 5
CNET - 93

27. Bellcore vs. 217 Recognition & Acceptance Concentration
Calculations & Equations
Consideration of Test Data
Multiplier
Parts
Environments
Quality Levels

28. Accuracy of MTBF Assessments
Stage I: Parts count method, assuming constant failure rates
Stage II: Variation of failure rates according to part families
Stage III: Taking into account of operational parameters
Stage IV: Consideration of failure modes, time influences, different failure distribution for each part, etc.
Accuracy
Time spent for the analysis

29. PRISM Reliability Model
Developed by the Reliability Analysis Center (RAC)
Accounts for the effect of process related variability on system failure rate
Inherent failure rate based on base failure rate and environmental conditions (RAC Rates model)
Failure rate may then be modified by:
Process Grade Factors, and/or
Bayesian Analysis, and/or
Predecessor Data

30. PRISM Methodology
Operational Profile, Environmental and Electrical Stresses
Process Assessments
RAC Component Models
Test Data
RAC Failure Rate Databases
System Reliability Assessment Model
Bayesian Data Combination
Historical Data on Similar Systems
System Reliability Estimate
Software Model

31. Primary Causes of Failure
(Nominal Values)

32. PRISM Process Grade Factor Types
Design
Manufacturing
Parts Quality
System Management
CND (Can Not Duplicate)
Induced
Wearout
Growth
Infant Mortality

33. Other PRISM Adjustments
Bayesian
Uses test and field data to enhance predicted failure rate
Predecessor
Uses previous history data to further refine predicted failure rate

34. PRISM Note
Although PRISM contains RAC Rate models for many part types, it does not include the following:
Rotating devices Relays
Switching devices Tubes
Connections Lasers
Miscellaneous parts
Relex can solve this problem by allowing the user to apply PRISM concepts (Process Grade, Bayesian, Predecessor) to a failure rate calculated by all other models.