Study guide
13.1 Reliability Modeling, Failure Rates (lambda) & Life Distributions (Exponential, Weibull)
PE Industrial and Systems Exam
Track your progress as you read
Finished sections get a check mark.
1 Exam Overview, Strategy & Systems Foundations
1.1 PE Industrial Exam Format, Official Specification & Reference Handbook Strategy 1.2 System Analysis Tools, Process Definition, Requirements Engineering & QFD 1.3 Performance Measures: Output, Productivity, OEE, Cost & Environmental Sustainability 1.4 Process Types (Discrete, Continuous, Manufacturing, Service) & Value Analysis/Engineering
2 Optimization, Queueing, Simulation, Networks & Markov Chains
2.1 Linear Programming Formulation, Graphical Solutions & Sensitivity Analysis 2.2 Queueing Theory Models (M/M/1, M/M/s) & Little's Law Applications 2.3 Discrete-Event Simulation, Random Variates & Model Validation 2.4 Graph Theory & Network Models: Shortest Path, Minimal Spanning Tree & Maximum Flow 2.5 Markov Chains: Transition Matrices, Steady-State Probabilities & Absorbing States 2.6 Decision Trees, Expected Value & Bottleneck Analysis (Theory of Constraints)
3 Project Management & Planning
3.1 Project Scheduling with CPM and PERT: Critical Path, Float, Probability & Crashing 3.2 Project Planning & Control: Scope and WBS, Gantt Charts, Critical Chain (CCPM), Risk & Earned Value
4 Facility Layout, Space, Capacity & Material Handling
4.1 Systematic Layout Planning (SLP) & Activity Relationship (REL) Analysis 4.2 Workstation Design, Aisle Sizing & Space Requirements Planning 4.3 Capacity Analysis: Machine Requirements, Scrap Allowances & Personnel Requirements 4.4 Material Handling Equipment Selection, Unit Loads & Handling Principles
5 Site Location, Warehousing & Facility Economics
5.1 Facility Location Models: Center of Gravity & Minisum / Minimax Distance 5.2 Warehousing Systems, Storage Assignment & Order Picking Operations 5.3 Facility Life-Cycle Economics, Replacement Analysis & Rearrangement Costs
6 Demand Forecasting, Production Planning, MRP & JIT
6.1 Demand Forecasting Methods: Moving Averages, Exponential Smoothing & Error Metrics 6.2 Aggregate Planning Strategies & Master Production Scheduling (MPS) 6.3 Material Requirements Planning (MRP), Bill of Materials & Capacity Planning 6.4 Just-in-Time & Lean Production Planning: Pull Systems, Kanban Sizing, Heijunka & SMED
7 Inventory Systems, Scheduling, Logistics & Transportation Modes
7.1 Deterministic Inventory Control: Classical EOQ, EPQ & Quantity Discounts 7.2 Stochastic Inventory Management: Safety Stock, Continuous / Periodic Review & Newsvendor 7.3 Production Sequencing, Dispatching Rules & Johnson's Algorithm 7.4 Supply Chain Logistics, Transportation Simplex & Distribution Network Design 7.5 Transportation Modes, Freight Economics & the Transshipment Model
8 Engineering Economics, Break-Even Analysis & Costing Systems
8.1 Equivalence, Cash Flow Analysis & Discounted Cash Flow Metrics (PW, FW, EAC) 8.2 Rate of Return Analysis (IRR, Incremental IRR) & Benefit-Cost (B/C) Ratios 8.3 Asset Depreciation (MACRS, Straight-Line), Tax Effects & Replacement Decisions 8.4 Break-Even Analysis, Technical Capability Assessment & Costing Systems (Traditional, ABC & Standard Costs)
9 Methods Engineering, Work Measurement, Learning Curves, Line Balancing & Process Planning
9.1 Methods Analysis, Process Charting & Motion Study (Therbligs, 5S) 9.2 Worker–Machine (Man–Machine) Charts, Multiple Activity Analysis & Synchronous Servicing 9.3 Time Study Fundamentals, Performance Rating & Allowance Factors 9.4 Predetermined Motion Time Systems (PMTS, MTM, MOST) & Work Sampling 9.5 Learning Curves: Unit and Cumulative-Average Models, Learning Rates & Applications 9.6 Assembly Line Balancing, Precedence Constraints & Efficiency Heuristics 9.7 Manufacturing & Service Process Planning: Operation Selection, Sequencing, Route Sheets, Fixturing & Mistake Proofing
10 Ergonomics, Human Capacity, Occupational Safety & DART Rates
10.1 Anthropometry, Percentile Design, Workstation Ergonomics & Link Analysis 10.2 Industrial Biomechanics, the NIOSH Lifting Equation & Lifting Aids 10.3 Limits of Human Capacity & Risk-Factor Tools: Work Physiology, Strength, Range of Motion, Vision, RULA & Slip Resistance 10.4 Environmental Ergonomics (Noise, Heat, Illumination), Human-Machine Systems & OSHA Safety 10.5 Safety Codes, Standards & Voluntary Guidelines (OSHA, ANSI, NIOSH, MIL-STD) and DART Rate Calculations
11 Statistical Process Control & Process Capability
11.1 Statistical Process Control for Variables (X-bar/R and X-bar/s Charts) 11.2 Statistical Process Control for Attributes (p, np, c, and u Charts) 11.3 Process Capability Indices (Cp, Cpk, Cpm, Pp, Ppk) & Sigma Levels
12 Design of Experiments, Acceptance Sampling & Quality Systems
12.1 Design of Experiments: 2^k Factorials, ANOVA & Taguchi Robust Design 12.2 Acceptance Sampling: OC Curves, AQL, LTPD, AOQL, Z1.4/MIL-STD-105E & Dodge-Romig 12.3 Quality Systems (Deming, TQM, ISO 9000, Benchmarking), Root Cause Analysis & DMAIC
13 Reliability Engineering & Maintenance Systems
13.1 Reliability Modeling, Failure Rates (lambda) & Life Distributions (Exponential, Weibull) 13.2 System Reliability Architecture: Series, Parallel, k-out-of-n & Redundancy 13.3 Equipment Maintenance Engineering, Availability (A), MTBF/MTTR & RCM
13.1 Reliability Modeling, Failure Rates (lambda) & Life Distributions (Exponential, Weibull)
Key Takeaways Explain This Topic AI
Reliability R ( t ) = P ( T > t ) = 1 − F ( t ) R(t) = P(T > t) = 1 - F(t) R ( t ) = P ( T > t ) = 1 − F ( t ) defines the probability that an asset performs without failure over mission duration t t t under specified operating conditions.
The instantaneous hazard rate h ( t ) = f ( t ) / R ( t ) = − d [ ln R ( t ) ] / d t h(t) = f(t)/R(t) = -d[\ln R(t)]/dt h ( t ) = f ( t ) / R ( t ) = − d [ ln R ( t )] / d t relates to reliability via R ( t ) = exp ( − ∫ 0 t h ( u ) d u ) R(t) = \exp(-\int_0^t h(u)\,du) R ( t ) = exp ( − ∫ 0 t h ( u ) d u ) , tracking failure intensity across the Bathtub Curve (infant mortality, useful life, wear-out).
The Exponential distribution models constant failure rates ( λ \lambda λ ) with memoryless property, where MTTF = 1 / λ \text{MTTF} = 1/\lambda MTTF = 1/ λ and R ( MTTF ) = e − 1 ≈ 36.79 % R(\text{MTTF}) = e^{-1} \approx 36.79\% R ( MTTF ) = e − 1 ≈ 36.79% , making time-based preventive replacement ineffective.
The Weibull distribution models flexible failure physics: shape β < 1 \beta < 1 β < 1 indicates infant mortality, β = 1 \beta = 1 β = 1 constant failure rate, and β > 1 \beta > 1 β > 1 wear-out, while scale parameter θ \theta θ represents the characteristic life (63.2% failure point).
Mechanical fatigue and micro-degradation follow Normal (additive wear) or Lognormal (multiplicative crack growth, electromigration) life distributions.
Last updated: October 2026
13.1 Reliability Modeling, Failure Rates (λ) & Life Distributions
Reliability engineering provides the quantitative framework for predicting, assessing, and optimizing the operational life cycle of components, production equipment, and complex industrial systems. In modern manufacturing and systems engineering, understanding failure probability over time is essential for establishing safety envelopes, warranty periods, spare parts provisioning, and condition-based maintenance policies.
1. Fundamentals of Reliability Mathematics
Let the operating time to failure of an engineering component or system be modeled as a continuous, non-negative random variable T ≥ 0 T \ge 0 T ≥ 0 , characterized by a probability density function (PDF) denoted by f ( t ) f(t) f ( t ) .
Reliability Function R ( t ) R(t) R ( t )
The Reliability function (also termed the Survival function S ( t ) S(t) S ( t ) ) defines the probability that the asset functions without failure beyond operating time t t t under specified operating and environmental conditions:
R ( t ) = P ( T > t ) = ∫ t ∞ f ( u ) d u R(t) = P(T > t) = \int_t^\infty f(u)\,du R ( t ) = P ( T > t ) = ∫ t ∞ f ( u ) d u
Boundary conditions require:
At time zero: R ( 0 ) = 1 R(0) = 1 R ( 0 ) = 1 (the asset is assumed operational at start of mission).
As time approaches infinity: lim t → ∞ R ( t ) = 0 \lim_{t \to \infty} R(t) = 0 lim t → ∞ R ( t ) = 0 (all physical systems eventually fail).
Monotonicity: d R ( t ) d t ≤ 0 \frac{dR(t)}{dt} \le 0 d t d R ( t ) ≤ 0 for all t ≥ 0 t \ge 0 t ≥ 0 .
Cumulative Failure Probability F ( t ) F(t) F ( t )
The Cumulative Failure Distribution function (or Unreliability function ) represents the complementary cumulative probability that the system fails at or before operating time t t t :
F ( t ) = P ( T ≤ t ) = 1 − R ( t ) = ∫ 0 t f ( u ) d u F(t) = P(T \le t) = 1 - R(t) = \int_0^t f(u)\,du F ( t ) = P ( T ≤ t ) = 1 − R ( t ) = ∫ 0 t f ( u ) d u
Probability Density Function f ( t ) f(t) f ( t )
The failure probability density function f ( t ) f(t) f ( t ) represents the instantaneous failure probability per unit time in the interval [ t , t + d t ] [t, t + dt] [ t , t + d t ] :
f ( t ) = d F ( t ) d t = − d R ( t ) d t f(t) = \frac{dF(t)}{dt} = -\frac{dR(t)}{dt} f ( t ) = d t d F ( t ) = − d t d R ( t )
Mean Time To Failure (MTTF)
For non-repairable assets, the expected operational life is the Mean Time To Failure (MTTF) . Applying integration by parts over the domain [ 0 , ∞ ) [0, \infty) [ 0 , ∞ ) :
MTTF = E [ T ] = ∫ 0 ∞ t f ( t ) d t = [ − t R ( t ) ] 0 ∞ + ∫ 0 ∞ R ( t ) d t \text{MTTF} = E[T] = \int_0^\infty t f(t)\,dt = \left[ -t R(t) \right]_0^\infty + \int_0^\infty R(t)\,dt MTTF = E [ T ] = ∫ 0 ∞ t f ( t ) d t = [ − tR ( t ) ] 0 ∞ + ∫ 0 ∞ R ( t ) d t
Assuming lim t → ∞ t R ( t ) = 0 \lim_{t \to \infty} t R(t) = 0 lim t → ∞ tR ( t ) = 0 , the standard identity simplifies to the integral of the reliability function:
MTTF = ∫ 0 ∞ R ( t ) d t \text{MTTF} = \int_0^\infty R(t)\,dt MTTF = ∫ 0 ∞ R ( t ) d t
For repairable systems undergoing renewal or replacement upon breakdown, this metric is expressed as the Mean Time Between Failures (MTBF) .
2. Hazard Rate and Instantaneous Failure Rate Function
A critical distinction in reliability theory exists between the absolute failure density f ( t ) f(t) f ( t ) and the conditional failure intensity given survival to time t t t . The hazard rate function h ( t ) h(t) h ( t ) (also designated instantaneous failure rate z ( t ) z(t) z ( t ) or λ ( t ) \lambda(t) λ ( t ) ) represents the conditional probability per unit time that an item fails in the infinitesimal interval ( t , t + Δ t ] (t, t + \Delta t] ( t , t + Δ t ] , given that it survived up to time t t t :
h ( t ) = lim Δ t → 0 P ( t < T ≤ t + Δ t ∣ T > t ) Δ t h(t) = \lim_{\Delta t \to 0} \frac{P(t < T \le t + \Delta t \mid T > t)}{\Delta t} h ( t ) = lim Δ t → 0 Δ t P ( t < T ≤ t + Δ t ∣ T > t )
Applying conditional probability P ( A ∣ B ) = P ( A ∩ B ) P ( B ) P(A \mid B) = \frac{P(A \cap B)}{P(B)} P ( A ∣ B ) = P ( B ) P ( A ∩ B ) :
P ( t < T ≤ t + Δ t ∣ T > t ) = F ( t + Δ t ) − F ( t ) R ( t ) P(t < T \le t + \Delta t \mid T > t) = \frac{F(t + \Delta t) - F(t)}{R(t)} P ( t < T ≤ t + Δ t ∣ T > t ) = R ( t ) F ( t + Δ t ) − F ( t )
Taking the limit as Δ t → 0 \Delta t \to 0 Δ t → 0 :
h ( t ) = 1 R ( t ) lim Δ t → 0 F ( t + Δ t ) − F ( t ) Δ t = f ( t ) R ( t ) = − 1 R ( t ) d R ( t ) d t = − d d t ln R ( t ) h(t) = \frac{1}{R(t)} \lim_{\Delta t \to 0} \frac{F(t + \Delta t) - F(t)}{\Delta t} = \frac{f(t)}{R(t)} = -\frac{1}{R(t)}\frac{dR(t)}{dt} = -\frac{d}{dt} \ln R(t) h ( t ) = R ( t ) 1 lim Δ t → 0 Δ t F ( t + Δ t ) − F ( t ) = R ( t ) f ( t ) = − R ( t ) 1 d t d R ( t ) = − d t d ln R ( t )
Derivation of the General Reliability Relationship
Integrating both sides of h ( u ) = − d d u ln R ( u ) h(u) = -\frac{d}{du} \ln R(u) h ( u ) = − d u d ln R ( u ) from 0 0 0 to t t t :
∫ 0 t h ( u ) d u = − ∫ 0 t d ( ln R ( u ) ) = − [ ln R ( t ) − ln R ( 0 ) ] \int_0^t h(u)\,du = -\int_0^t d(\ln R(u)) = -\left[\ln R(t) - \ln R(0)\right] ∫ 0 t h ( u ) d u = − ∫ 0 t d ( ln R ( u )) = − [ ln R ( t ) − ln R ( 0 ) ]
Because R ( 0 ) = 1 R(0) = 1 R ( 0 ) = 1 , ln R ( 0 ) = 0 \ln R(0) = 0 ln R ( 0 ) = 0 . Defining the Cumulative Hazard Function as H ( t ) = ∫ 0 t h ( u ) d u H(t) = \int_0^t h(u)\,du H ( t ) = ∫ 0 t h ( u ) d u :
ln R ( t ) = − ∫ 0 t h ( u ) d u = − H ( t ) \ln R(t) = -\int_0^t h(u)\,du = -H(t) ln R ( t ) = − ∫ 0 t h ( u ) d u = − H ( t )
Exponentiating both sides yields the foundational equation connecting reliability directly to the hazard rate:
R ( t ) = exp ( − ∫ 0 t h ( u ) d u ) = e − H ( t ) R(t) = \exp\left( -\int_0^t h(u)\,du \right) = e^{-H(t)} R ( t ) = exp ( − ∫ 0 t h ( u ) d u ) = e − H ( t )
Consequently, the failure density function can be expressed purely in terms of the hazard rate:
f ( t ) = h ( t ) exp ( − ∫ 0 t h ( u ) d u ) = h ( t ) e − H ( t ) f(t) = h(t) \exp\left( -\int_0^t h(u)\,du \right) = h(t) e^{-H(t)} f ( t ) = h ( t ) exp ( − ∫ 0 t h ( u ) d u ) = h ( t ) e − H ( t )
3. The Classic Bathtub Curve
In physical and mechanical systems, the instantaneous hazard rate h ( t ) h(t) h ( t ) across an asset's full operating life typically exhibits three distinct chronological regimes, forming the classic Bathtub Curve .
Phase Lifecycle Region Hazard Slope Primary Failure Drivers Optimal Engineering Countermeasures
Phase I Infant Mortality (Early Failure / Burn-in) d h ( t ) d t < 0 \frac{dh(t)}{dt} < 0 d t d h ( t ) < 0 (Decreasing Failure Rate, DFR) Manufacturing defects, sub-standard materials, assembly misalignments, poor solder joints, contamination Environmental Stress Screening (ESS), accelerated thermal/vibration burn-in, rigorous quality acceptance inspection
Phase II Useful Life (Normal Operating Period) d h ( t ) d t = 0 ⟹ h ( t ) = λ = constant \frac{dh(t)}{dt} = 0 \implies h(t) = \lambda = \text{constant} d t d h ( t ) = 0 ⟹ h ( t ) = λ = constant Random environmental shocks, external voltage spikes, sudden overload transients, stress-strength interference Redundant system architecture, component derating, safety margins, robust fail-safe engineering
Phase III Wear-Out (Aging / End of Life) d h ( t ) d t > 0 \frac{dh(t)}{dt} > 0 d t d h ( t ) > 0 (Increasing Failure Rate, IFR) Mechanical fatigue, progressive friction wear, oxidation, corrosion, creep, dielectric breakdown Predictive condition monitoring (vibration, oil analysis), scheduled preventive replacement before wear acceleration Hazard Rate h(t)
- ^
- |
- | \ /
- | \ /
- | \______________________________________/
- | <-- Infant --> <--- Useful Life ---> <-- Wear-out -->
- | Mortality (DFR) Constant (CFR) (IFR)
- +----------------------------------------------------> Time (t)
- Strategic Engineering Implications
During Phase I , scheduled preventive maintenance is counterproductive. Disassembling assets experiencing early infant mortality introduces human maintenance-induced errors, resetting components into infant mortality failure risk.
During Phase II , the asset does not experience age-related degradation. The failure probability in any subsequent hour is entirely independent of prior operational service time.
During Phase III , failure intensity escalates rapidly. Timely scheduled overhaul or component retirement based on cumulative operating cycles or run hours prevents catastrophic in-service breakdowns.
4. The Exponential Distribution (Constant Hazard Rate)
The Exponential Distribution represents the foundational life distribution in reliability modeling, characterizing assets operating in Phase II of the bathtub curve where failure is governed strictly by random, Poisson-distributed shocks.
Mathematical Formulation
Setting the hazard rate to a time-invariant constant h ( t ) = λ > 0 h(t) = \lambda > 0 h ( t ) = λ > 0 :
- H ( t ) = ∫ 0 t λ d u = λ t H(t) = \int_0^t \lambda\,du = \lambda t H ( t ) = ∫ 0 t λ d u = λ t
- R ( t ) = e − λ t R(t) = e^{-\lambda t} R ( t ) = e − λ t
- F ( t ) = 1 − e − λ t F(t) = 1 - e^{-\lambda t} F ( t ) = 1 − e − λ t
- f ( t ) = λ e − λ t f(t) = \lambda e^{-\lambda t} f ( t ) = λ e − λ t
- Expected Life and Variance
MTTF = ∫ 0 ∞ e − λ t d t = [ − 1 λ e − λ t ] 0 ∞ = 1 λ \text{MTTF} = \int_0^\infty e^{-\lambda t}\,dt = \left[ -\frac{1}{\lambda} e^{-\lambda t} \right]_0^\infty = \frac{1}{\lambda} MTTF = ∫ 0 ∞ e − λ t d t = [ − λ 1 e − λ t ] 0 ∞ = λ 1
Var ( T ) = 1 λ 2 , σ = 1 λ \text{Var}(T) = \frac{1}{\lambda^2}, \quad \sigma = \frac{1}{\lambda} Var ( T ) = λ 2 1 , σ = λ 1
The Memoryless Property
A continuous random variable T T T is memoryless if and only if for all t > 0 t > 0 t > 0 and s > 0 s > 0 s > 0 :
P ( T > t + s ∣ T > s ) = P ( T > t + s ) P ( T > s ) = e − λ ( t + s ) e − λ s = e − λ t = P ( T > t ) P(T > t + s \mid T > s) = \frac{P(T > t + s)}{P(T > s)} = \frac{e^{-\lambda(t + s)}}{e^{-\lambda s}} = e^{-\lambda t} = P(T > t) P ( T > t + s ∣ T > s ) = P ( T > s ) P ( T > t + s ) = e − λ s e − λ ( t + s ) = e − λ t = P ( T > t )
Physical Meaning: A component operating under a constant failure rate that has run without failure for s = 5,000 s = 5{,}000 s = 5 , 000 hours has the exact same probability of surviving the next t = 1,000 t = 1{,}000 t = 1 , 000 hours as a brand new component fresh off the production line. Wear, usage, and age do not accumulate. Consequently, replacing unfailed exponential components based on calendar time or operational hours wastes capital and provides zero reliability benefit.
Reliability at t = MTTF t = \text{MTTF} t = MTTF
Evaluating the reliability function at an operating time equal to the mean life t = 1 / λ t = 1/\lambda t = 1/ λ :
R ( MTTF ) = e − λ ( 1 / λ ) = e − 1 = 1 2.71828 ≈ 0.367879 ( 36.79 % ) R(\text{MTTF}) = e^{-\lambda(1/\lambda)} = e^{-1} = \frac{1}{2.71828} \approx 0.367879 \quad (36.79\%) R ( MTTF ) = e − λ ( 1/ λ ) = e − 1 = 2.71828 1 ≈ 0.367879 ( 36.79% )
Key Takeaway: An asset governed by an exponential life distribution has only a 36.8% chance of surviving its full MTTF. Conversely, 63.2% of such components fail before reaching MTTF.
5. The Weibull Distribution (Time-Dependent Hazard)
Developed by Waloddi Weibull in 1951, the Weibull distribution is the most versatile parametric distribution in reliability engineering because its hazard function can model decreasing, constant, or increasing failure rates depending on the value of its shape parameter.
Two-Parameter Weibull Equations
The two-parameter Weibull reliability function is parameterized by shape parameter β > 0 \beta > 0 β > 0 (dimensionless) and scale parameter θ > 0 \theta > 0 θ > 0 (characteristic life, having units of time):
R ( t ) = exp ( − ( t θ ) β ) R(t) = \exp\left( -\left(\frac{t}{\theta}\right)^\beta \right) R ( t ) = exp ( − ( θ t ) β )
F ( t ) = 1 − exp ( − ( t θ ) β ) F(t) = 1 - \exp\left( -\left(\frac{t}{\theta}\right)^\beta \right) F ( t ) = 1 − exp ( − ( θ t ) β )
Taking the derivative gives the probability density function f ( t ) f(t) f ( t ) :
f ( t ) = − d R ( t ) d t = β θ ( t θ ) β − 1 exp ( − ( t θ ) β ) f(t) = -\frac{dR(t)}{dt} = \frac{\beta}{\theta} \left( \frac{t}{\theta} \right)^{\beta - 1} \exp\left( -\left(\frac{t}{\theta}\right)^\beta \right) f ( t ) = − d t d R ( t ) = θ β ( θ t ) β − 1 exp ( − ( θ t ) β )
The instantaneous hazard rate h ( t ) = f ( t ) / R ( t ) h(t) = f(t)/R(t) h ( t ) = f ( t ) / R ( t ) simplifies directly to:
h ( t ) = β θ ( t θ ) β − 1 h(t) = \frac{\beta}{\theta} \left( \frac{t}{\theta} \right)^{\beta - 1} h ( t ) = θ β ( θ t ) β − 1
Physical Interpretation of the Shape Parameter β \beta β
The exponent ( β − 1 ) (\beta - 1) ( β − 1 ) in h ( t ) h(t) h ( t ) directly governs the slope of the hazard function:
β < 1 \beta < 1 β < 1 (Decreasing Failure Rate, DFR): Infant mortality regime. Hazard rate declines over time ( h ′ ( t ) < 0 h'(t) < 0 h ′ ( t ) < 0 ). Assets weed out early defects. Burn-in screening is advantageous.
β = 1 \beta = 1 β = 1 (Constant Failure Rate, CFR): Hazard rate reduces to h ( t ) = 1 θ = λ h(t) = \frac{1}{\theta} = \lambda h ( t ) = θ 1 = λ . The Weibull reduces identically to the Exponential distribution with MTTF = θ \text{MTTF} = \theta MTTF = θ .
β > 1 \beta > 1 β > 1 (Increasing Failure Rate, IFR): Progressive wear-out regime ( h ′ ( t ) > 0 h'(t) > 0 h ′ ( t ) > 0 ). Assets degrade with operating cycles, friction, fatigue, or aging. Preventive replacement strategies are highly effective.
β = 2 \beta = 2 β = 2 : Rayleigh distribution, where the hazard rate increases linearly with time: h ( t ) = 2 t θ 2 h(t) = \frac{2t}{\theta^2} h ( t ) = θ 2 2 t .
β ≈ 3.44 \beta \approx 3.44 β ≈ 3.44 : The Weibull distribution exhibits zero skewness, closely approximating a symmetrical Gaussian (Normal) distribution with mean μ ≈ 0.90 θ \mu \approx 0.90\theta μ ≈ 0.90 θ and standard deviation σ ≈ 0.29 θ \sigma \approx 0.29\theta σ ≈ 0.29 θ .
β ≥ 4 \beta \ge 4 β ≥ 4 : High wear-out steepness, characteristic of severe mechanical fatigue, rapid bearing degradation, or thermal embrittlement.
Characteristic Life θ \theta θ
Evaluating the Weibull reliability function at t = θ t = \theta t = θ :
R ( θ ) = exp ( − ( θ θ ) β ) = e − 1 β = e − 1 ≈ 0.367879 ( 36.79 % ) R(\theta) = \exp\left( -\left(\frac{\theta}{\theta}\right)^\beta \right) = e^{-1^\beta} = e^{-1} \approx 0.367879 \quad (36.79\%) R ( θ ) = exp ( − ( θ θ ) β ) = e − 1 β = e − 1 ≈ 0.367879 ( 36.79% )
Regardless of the value of shape parameter β \beta β , the scale parameter θ \theta θ is always the operational age at which 63.2% of the asset population has failed and 36.8% remains operational. For this reason, θ \theta θ is formally defined as the Characteristic Life .
Mean Time To Failure for the Weibull Model
The expected life involves the Euler Gamma function Γ ( x ) = ∫ 0 ∞ u x − 1 e − u d u \Gamma(x) = \int_0^\infty u^{x-1} e^{-u}\,du Γ ( x ) = ∫ 0 ∞ u x − 1 e − u d u :
MTTF = θ Γ ( 1 + 1 β ) \text{MTTF} = \theta\,\Gamma\left( 1 + \frac{1}{\beta} \right) MTTF = θ Γ ( 1 + β 1 )
Var ( T ) = θ 2 [ Γ ( 1 + 2 β ) − ( Γ ( 1 + 1 β ) ) 2 ] \text{Var}(T) = \theta^2 \left[ \Gamma\left( 1 + \frac{2}{\beta} \right) - \left( \Gamma\left( 1 + \frac{1}{\beta} \right) \right)^2 \right] Var ( T ) = θ 2 [ Γ ( 1 + β 2 ) − ( Γ ( 1 + β 1 ) ) 2 ]
Three-Parameter Weibull (Guaranteed Operating Life)
When an asset exhibits zero probability of failure prior to a threshold operating duration γ > 0 \gamma > 0 γ > 0 (the location parameter or failure-free life), the distribution shifts rightward for t ≥ γ t \ge \gamma t ≥ γ :
R ( t ) = exp ( − ( t − γ θ ) β ) , t ≥ γ R(t) = \exp\left( -\left(\frac{t - \gamma}{\theta}\right)^\beta \right), \quad t \ge \gamma R ( t ) = exp ( − ( θ t − γ ) β ) , t ≥ γ
h ( t ) = β θ ( t − γ θ ) β − 1 , t ≥ γ h(t) = \frac{\beta}{\theta} \left( \frac{t - \gamma}{\theta} \right)^{\beta - 1}, \quad t \ge \gamma h ( t ) = θ β ( θ t − γ ) β − 1 , t ≥ γ
6. Normal and Lognormal Life Distributions
While the Exponential and Weibull distributions dominate standard reliability modeling, structural fatigue and electronic degradation frequently adhere to Normal and Lognormal behaviors.
Normal (Gaussian) Life Distribution
The Normal distribution models degradation where total wear results from the additive accumulation of many small, independent physical increments (Central Limit Theorem):
f ( t ) = 1 σ 2 π exp ( − ( t − μ ) 2 2 σ 2 ) f(t) = \frac{1}{\sigma \sqrt{2\pi}} \exp\left( -\frac{(t - \mu)^2}{2\sigma^2} \right) f ( t ) = σ 2 π 1 exp ( − 2 σ 2 ( t − μ ) 2 )
R ( t ) = 1 − Φ ( t − μ σ ) R(t) = 1 - \Phi\left( \frac{t - \mu}{\sigma} \right) R ( t ) = 1 − Φ ( σ t − μ )
Where Φ ( z ) \Phi(z) Φ ( z ) is the standard normal cumulative distribution function. The hazard rate h ( t ) = f ( t ) / R ( t ) h(t) = f(t)/R(t) h ( t ) = f ( t ) / R ( t ) for a Normal distribution increases monotonically with operating time ( h ′ ( t ) > 0 h'(t) > 0 h ′ ( t ) > 0 ), making it suitable for mechanical structural wear-out with fixed service lives.
Lognormal Life Distribution
The Lognormal distribution governs failure processes where degradation is multiplicative rather than additive. If the random variable T T T is lognormally distributed, then Y = ln ( T ) Y = \ln(T) Y = ln ( T ) is normally distributed: Y ∼ N ( μ ln , σ ln 2 ) Y \sim N(\mu_{\ln}, \sigma_{\ln}^2) Y ∼ N ( μ l n , σ l n 2 ) .
f ( t ) = 1 t σ ln 2 π exp ( − ( ln t − μ ln ) 2 2 σ ln 2 ) , t > 0 f(t) = \frac{1}{t \sigma_{\ln} \sqrt{2\pi}} \exp\left( -\frac{(\ln t - \mu_{\ln})^2}{2\sigma_{\ln}^2} \right), \quad t > 0 f ( t ) = t σ l n 2 π 1 exp ( − 2 σ l n 2 ( l n t − μ l n ) 2 ) , t > 0
R ( t ) = 1 − Φ ( ln t − μ ln σ ln ) R(t) = 1 - \Phi\left( \frac{\ln t - \mu_{\ln}}{\sigma_{\ln}} \right) R ( t ) = 1 − Φ ( σ l n l n t − μ l n )
Engineering Applications of the Lognormal Distribution
Fatigue crack growth: Linear elastic fracture mechanics (Paris' Law) exhibits exponential crack propagation rates d a / d N = C ( Δ K ) m da/dN = C(\Delta K)^m d a / d N = C ( Δ K ) m , leading to multiplicative crack enlargement.
Semiconductor physics: Electromigration in metallic interconnects (governed by Black's Equation), dielectric breakdown, and corrosion kinetics.
Hazard rate profile: Unlike the Weibull IFR, the Lognormal hazard rate starts at zero, climbs to a peak, and then decreases asymptotically toward zero at very large t t t . The physical rationale is that surviving units past the peak represent an exceptionally robust subpopulation.
- 7. Worked Numerical Example: Comparative Life Analytics
- Problem Formulation
A high-reliability variable-speed feedwater pump installed in an industrial thermal plant has a characteristic life of θ = 8,000 \theta = 8{,}000 θ = 8 , 000 operating hours. A reliability engineering team must evaluate the pump's survival metrics over an operating mission of t 1 = 2,000 t_1 = 2{,}000 t 1 = 2 , 000 hours and an extended operating cycle of t 2 = 10,000 t_2 = 10{,}000 t 2 = 10 , 000 hours under two competing lifecycle models:
Model A (Exponential Model): Assumes constant failure rate throughout life with MTTF = 8,000 \text{MTTF} = 8{,}000 MTTF = 8 , 000 hours, so λ = 1 / 8,000 = 1.25 × 10 − 4 failures/hour \lambda = 1/8{,}000 = 1.25 \times 10^{-4}\text{ failures/hour} λ = 1/8 , 000 = 1.25 × 1 0 − 4 failures/hour .
Model B (Weibull Wear-Out Model): Assumes mechanical bearing wear-out with shape parameter β = 2.5 \beta = 2.5 β = 2.5 and scale parameter θ = 8,000 \theta = 8{,}000 θ = 8 , 000 hours.
Determine for both models:
The reliability R ( t ) R(t) R ( t ) at t = 2,000 t = 2{,}000 t = 2 , 000 hr and t = 10,000 t = 10{,}000 t = 10 , 000 hr.
The cumulative failure probability F ( t ) F(t) F ( t ) at both horizons.
The instantaneous hazard rate h ( t ) h(t) h ( t ) at both horizons.
Step-by-Step Solution
Part 1: Model A (Exponential Distribution, λ = 1.25 × 10 − 4 hr − 1 \lambda = 1.25 \times 10^{-4}\text{ hr}^{-1} λ = 1.25 × 1 0 − 4 hr − 1 )
At t 1 = 2,000 t_1 = 2{,}000 t 1 = 2 , 000 hours:
Operating exponent: λ t 1 = ( 1.25 × 10 − 4 hr − 1 ) ( 2,000 hr ) = 0.250 \lambda t_1 = (1.25 \times 10^{-4}\text{ hr}^{-1})(2{,}000\text{ hr}) = 0.250 λ t 1 = ( 1.25 × 1 0 − 4 hr − 1 ) ( 2 , 000 hr ) = 0.250
Reliability: R A ( 2,000 ) = e − 0.250 = 0.778801 ⟹ 77.88 % R_A(2{,}000) = e^{-0.250} = 0.778801 \implies 77.88\% R A ( 2 , 000 ) = e − 0.250 = 0.778801 ⟹ 77.88%
Cumulative failure probability: F A ( 2,000 ) = 1 − 0.778801 = 0.221199 ⟹ 22.12 % F_A(2{,}000) = 1 - 0.778801 = 0.221199 \implies 22.12\% F A ( 2 , 000 ) = 1 − 0.778801 = 0.221199 ⟹ 22.12%
Instantaneous hazard rate (constant for all t t t ): h A ( 2,000 ) = λ = 1.250 × 10 − 4 failures/hour h_A(2{,}000) = \lambda = 1.250 \times 10^{-4}\text{ failures/hour} h A ( 2 , 000 ) = λ = 1.250 × 1 0 − 4 failures/hour
At t 2 = 10,000 t_2 = 10{,}000 t 2 = 10 , 000 hours:
Operating exponent: λ t 2 = ( 1.25 × 10 − 4 hr − 1 ) ( 10,000 hr ) = 1.250 \lambda t_2 = (1.25 \times 10^{-4}\text{ hr}^{-1})(10{,}000\text{ hr}) = 1.250 λ t 2 = ( 1.25 × 1 0 − 4 hr − 1 ) ( 10 , 000 hr ) = 1.250
Reliability: R A ( 10,000 ) = e − 1.250 = 0.286505 ⟹ 28.65 % R_A(10{,}000) = e^{-1.250} = 0.286505 \implies 28.65\% R A ( 10 , 000 ) = e − 1.250 = 0.286505 ⟹ 28.65%
Cumulative failure probability: F A ( 10,000 ) = 1 − 0.286505 = 0.713495 ⟹ 71.35 % F_A(10{,}000) = 1 - 0.286505 = 0.713495 \implies 71.35\% F A ( 10 , 000 ) = 1 − 0.286505 = 0.713495 ⟹ 71.35%
Instantaneous hazard rate: h A ( 10,000 ) = λ = 1.250 × 10 − 4 failures/hour h_A(10{,}000) = \lambda = 1.250 \times 10^{-4}\text{ failures/hour} h A ( 10 , 000 ) = λ = 1.250 × 1 0 − 4 failures/hour
Part 2: Model B (Weibull Distribution, β = 2.5 , θ = 8,000 hr \beta = 2.5, \theta = 8{,}000\text{ hr} β = 2.5 , θ = 8 , 000 hr )
At t 1 = 2,000 t_1 = 2{,}000 t 1 = 2 , 000 hours:
Normalized time ratio: t 1 θ = 2,000 8,000 = 0.250 \frac{t_1}{\theta} = \frac{2{,}000}{8{,}000} = 0.250 θ t 1 = 8 , 000 2 , 000 = 0.250
Power term: ( t 1 θ ) β = ( 0.250 ) 2.5 = ( 0.250 ) 2 × 0.250 = 0.0625 × 0.500 = 0.03125 \left(\frac{t_1}{\theta}\right)^\beta = (0.250)^{2.5} = (0.250)^2 \times \sqrt{0.250} = 0.0625 \times 0.500 = 0.03125 ( θ t 1 ) β = ( 0.250 ) 2.5 = ( 0.250 ) 2 × 0.250 = 0.0625 × 0.500 = 0.03125
Reliability: R B ( 2,000 ) = exp ( − 0.03125 ) = e − 0.03125 = 0.969233 ⟹ 96.92 % R_B(2{,}000) = \exp(-0.03125) = e^{-0.03125} = 0.969233 \implies 96.92\% R B ( 2 , 000 ) = exp ( − 0.03125 ) = e − 0.03125 = 0.969233 ⟹ 96.92%
Cumulative failure probability: F B ( 2,000 ) = 1 − 0.969233 = 0.030767 ⟹ 3.08 % F_B(2{,}000) = 1 - 0.969233 = 0.030767 \implies 3.08\% F B ( 2 , 000 ) = 1 − 0.969233 = 0.030767 ⟹ 3.08%
Instantaneous hazard rate: h B ( t ) = β θ ( t θ ) β − 1 h_B(t) = \frac{\beta}{\theta} \left( \frac{t}{\theta} \right)^{\beta - 1} h B ( t ) = θ β ( θ t ) β − 1 β θ = 2.5 8,000 = 3.125 × 10 − 4 hr − 1 \frac{\beta}{\theta} = \frac{2.5}{8{,}000} = 3.125 \times 10^{-4}\text{ hr}^{-1} θ β = 8 , 000 2.5 = 3.125 × 1 0 − 4 hr − 1 ( t 1 θ ) β − 1 = ( 0.250 ) 1.5 = 0.250 × 0.250 = 0.125 \left(\frac{t_1}{\theta}\right)^{\beta - 1} = (0.250)^{1.5} = 0.250 \times \sqrt{0.250} = 0.125 ( θ t 1 ) β − 1 = ( 0.250 ) 1.5 = 0.250 × 0.250 = 0.125 h B ( 2,000 ) = ( 3.125 × 10 − 4 ) ( 0.125 ) = 3.90625 × 10 − 5 failures/hour h_B(2{,}000) = (3.125 \times 10^{-4})(0.125) = 3.90625 \times 10^{-5}\text{ failures/hour} h B ( 2 , 000 ) = ( 3.125 × 1 0 − 4 ) ( 0.125 ) = 3.90625 × 1 0 − 5 failures/hour
At t 2 = 10,000 t_2 = 10{,}000 t 2 = 10 , 000 hours:
Normalized time ratio: t 2 θ = 10,000 8,000 = 1.250 \frac{t_2}{\theta} = \frac{10{,}000}{8{,}000} = 1.250 θ t 2 = 8 , 000 10 , 000 = 1.250
Power term: ( t 2 θ ) β = ( 1.250 ) 2.5 = ( 1.250 ) 2 × 1.250 = 1.5625 × 1.118034 = 1.746928 \left(\frac{t_2}{\theta}\right)^\beta = (1.250)^{2.5} = (1.250)^2 \times \sqrt{1.250} = 1.5625 \times 1.118034 = 1.746928 ( θ t 2 ) β = ( 1.250 ) 2.5 = ( 1.250 ) 2 × 1.250 = 1.5625 × 1.118034 = 1.746928
Reliability: R B ( 10,000 ) = exp ( − 1.746928 ) = e − 1.746928 = 0.174308 ⟹ 17.43 % R_B(10{,}000) = \exp(-1.746928) = e^{-1.746928} = 0.174308 \implies 17.43\% R B ( 10 , 000 ) = exp ( − 1.746928 ) = e − 1.746928 = 0.174308 ⟹ 17.43%
Cumulative failure probability: F B ( 10,000 ) = 1 − 0.174308 = 0.825692 ⟹ 82.57 % F_B(10{,}000) = 1 - 0.174308 = 0.825692 \implies 82.57\% F B ( 10 , 000 ) = 1 − 0.174308 = 0.825692 ⟹ 82.57%
Instantaneous hazard rate: ( t 2 θ ) β − 1 = ( 1.250 ) 1.5 = 1.250 × 1.118034 = 1.397542 \left(\frac{t_2}{\theta}\right)^{\beta - 1} = (1.250)^{1.5} = 1.250 \times 1.118034 = 1.397542 ( θ t 2 ) β − 1 = ( 1.250 ) 1.5 = 1.250 × 1.118034 = 1.397542 h B ( 10,000 ) = ( 3.125 × 10 − 4 ) ( 1.397542 ) = 4.36732 × 10 − 4 failures/hour h_B(10{,}000) = (3.125 \times 10^{-4})(1.397542) = 4.36732 \times 10^{-4}\text{ failures/hour} h B ( 10 , 000 ) = ( 3.125 × 1 0 − 4 ) ( 1.397542 ) = 4.36732 × 1 0 − 4 failures/hour
Comparative Engineering Synthesis Table
Operating Metric Horizon: 2,000 Hours (Early Life) Horizon: 2,000 Hours Horizon: 10,000 Hours (Extended Life) Horizon: 10,000 Hours
Model Model A (Exponential, CFR) Model B (Weibull, IFR β=2.5) Model A (Exponential, CFR) Model B (Weibull, IFR β=2.5)
- Reliability R ( t ) R(t) R ( t ) 77.88% 96.92% 28.65% 17.43%
- Failure Probability F ( t ) F(t) F ( t ) 22.12% 3.08% 71.35% 82.57%
Hazard Rate h ( t ) h(t) h ( t ) 1.250 × 10 − 4 hr − 1 1.250 \times 10^{-4}\text{ hr}^{-1} 1.250 × 1 0 − 4 hr − 1 0.391 × 10 − 4 hr − 1 0.391 \times 10^{-4}\text{ hr}^{-1} 0.391 × 1 0 − 4 hr − 1 1.250 × 10 − 4 hr − 1 1.250 \times 10^{-4}\text{ hr}^{-1} 1.250 × 1 0 − 4 hr − 1 4.367 × 10 − 4 hr − 1 4.367 \times 10^{-4}\text{ hr}^{-1} 4.367 × 1 0 − 4 hr − 1
Critical Engineering Takeaway
At t = 2,000 t = 2{,}000 t = 2 , 000 hours, assuming an exponential distribution drastically underestimates the true reliability of a mechanical component ( 77.88 % 77.88\% 77.88% vs 96.92 % 96.92\% 96.92% ) because it presumes the component has been subject to full random failure intensity from time zero. Conversely, at t = 10,000 t = 10{,}000 t = 10 , 000 hours, the exponential model dangerously overestimates reliability ( 28.65 % 28.65\% 28.65% vs 17.43 % 17.43\% 17.43% ) and underestimates the failure hazard rate by a factor of 3.5 ( 1.25 × 10 − 4 1.25 \times 10^{-4} 1.25 × 1 0 − 4 vs 4.37 × 10 − 4 hr − 1 4.37 \times 10^{-4}\text{ hr}^{-1} 4.37 × 1 0 − 4 hr − 1 ) because it ignores progressive wear-out.