TM 5-698-3, Reliability Primer for Command, Control, Communications, Computer, Intelligence, Surveillance, and Reconnaissance (C4ISR) Facilities (14 October 2005)

Overview

TM 5-698-3 gives personnel who acquire and support equipment for mission-critical facilities such as command and data processing centres a basic introduction to and overview of reliability. The theory is purposely limited and kept simple, drawing on commercial practice and lessons learned from many reliability programmes. It supersedes the edition of 10 July 2003. It notes that reliability has grown in importance over the past 30 years as systems became more complex, support costs rose and defence budgets fell.

Chapter 2 gives the mathematical foundation: probability and statistics, reliability functions, sampling and estimation. For the exponential distribution reliability is R(t) = e^(-lambda t) and MTBF is 1/lambda. Chapter 3 covers reliability prediction and block-diagram modelling; the authors remark that the word 'prediction' sounds like fortune telling when it is only a quantitative assessment of the reliability inherent in a design or achieved in a product. A series system's reliability is the product of its elements; for active parallel, R = 1 - (1-R1)(1-R2)(1-R3).

Chapter 4 treats designing for reliability in eight steps: establish and allocate requirements, develop a system reliability model, conduct analyses (FMECA, fault tree, thermal analysis, worst-case circuit analysis and others), design, development testing, iterate the design, demonstration tests and a team approach. It carries an important warning on redundancy: redundancy raises only system-level mission reliability, not the reliability of individual items, and in fact lowers basic reliability because more failures occur, even if they are not mission failures, which raises support costs. Chapter 5 covers production and quality control and chapter 6 data collection and trending in operation.

The limit is that the manual is general and tutorial; it contains no facility-specific failure-rate data, which sits in other volumes of the TM 5-698 series (notably TM 5-698-5). It dates from 2005. Appendix B covers inherent and operational availability, and Appendix C gives examples of failure rate, similarity analysis, stress-strength interference, empirical models and failure data analysis. For a data centre designer the most useful parts are the block-diagram formulas and the warnings on redundancy.

Key points

1. Under an exponential failure distribution reliability is R(t) = e^(-lambda t), and MTBF (or MTTF) equals 1/lambda.
2. Series system reliability is the product of element reliabilities: R = R1 x R2 x R3.
3. Active parallel redundancy: R_system = 1 - (1-R1)(1-R2)(1-R3); a three-element unpowered standby set with perfect switching gives R = e^(-lambda t) + lambda t e^(-lambda t) + 0.5 (lambda t)^2 e^(-lambda t).
4. Allocation example: for 95 percent system reliability over a 10-hour mission, subsystems A, B and C must satisfy RA x RB x RC = 0.95.
5. Inherent availability Ai = MTBF / (MTBF + MTTR) excludes logistic and administrative delays.
6. Operational availability Ao = MTBM / (MTBM + MDT) x 100 percent accounts for all preventive and corrective maintenance and logistics delays; a maintainability specification such as MMax(95) = 6 h means 95 percent of repairs finish within 6 hours.
7. Redundancy lowers basic reliability: more components mean more failures and repairs, plus added weight, cost and complexity.
8. Reliability usually degrades over time; causes include changes in operating concept or environment, inadequate training, wearout and supplier changes, and the true root cause must be found before the design is changed.
9. Standard parts should be preferred and applied conservatively (derating); using a part at maximum capability raises its failure rate and leaves no margin for transients.

Using it in design

• Draw the UPS, generator and chiller chain as a series-parallel block diagram and compute system reliability on paper first, with the product rule for series parts and the parallel formula for redundant ones.
• When allocating a system reliability target to subsystems, check the product condition (RA x RB x RC = 0.95 in the example).
• Before adding redundancy, weigh the extra failures, maintenance load and cost of the added components; redundancy raises mission reliability but lowers basic reliability.
• Define the availability target in both inherent (MTBF/MTTR) and operational (MTBM/MDT) form, and write the maintainability requirement as a measurable condition such as a 95 percent repair time.
• Add derating of equipment below its rated limit and environmental characterisation (temperature, vibration and others) at design stage to the design criteria.
• Collect failure data in operation and trend it; when reliability falls, examine training, operating conditions and maintenance first before turning to design changes.

Who should read it

Newcomers to reliability engineering and designers who will assess data centre topologies with availability mathematics.

Document contents

1. Chapter 1 Introduction to Reliability: purpose, scope, definitions, historical perspective, importance of reliability
2. Chapter 2 Reliability and Its Mathematical Foundations
3. 2-2 Mathematical foundations: probability and statistics
4. 2-4 Sampling and estimation
5. Chapter 3 Reliability Prediction
6. 3-5 Reliability modeling (series, parallel, k of N)
7. Chapter 4 Designing for Reliability
8. 4-1 Establish and allocate requirements
9. 4-3 Conduct analyses
10. 4-4 Design for reliability
11. 4-5 Conduct development testing
12. 4-7 Conduct demonstration tests
13. Chapter 5 Producing Reliable Systems
14. Chapter 6 Reliability Improvement: data collection, trending, corrective actions
15. Appendix B Availability
16. Appendix C Examples of Reliability Prediction Methods

Read the document

Publication details

Authors

Headquarters, Department of the Army

Publisher

U.S. Army Corps of Engineers (Headquarters, Department of the Army technical manual)

Year

2005

Document language

English

Source type

Government / public

Access

Open licence — downloadable

Licence

U.S. public domain — U.S. Government work; free to reuse

Licence evidence

www.copyright.gov

Citation