23.1 Reliability Mathematics, Hazard Rates, and the Bathtub Curve

Reliability engineering provides the quantitative framework for predicting, assessing, and optimizing the operational life cycle of industrial components and systems. In the NCEES FE Industrial and Systems examination, reliability problems frequently test fundamental probability relationships, failure rate conversions, component survival over specific mission times, and the operational implications of the classical bathtub curve.

---

1. Fundamental Mathematics of Reliability

Reliability modeling treats the time-to-failure of a component as a continuous, non-negative random variable denoted by $T$ ($T \ge 0$). Four interrelated mathematical functions govern reliability analysis: the cumulative failure distribution $F(t)$, the reliability function $R(t)$, the failure probability density function $f(t)$, and the hazard function $h(t)$.

                  Interrelationships of Reliability Functions

                             ┌─────────────────┐
                             │  PDF:  f(t)     │
                             └────────┬────────┘
                                      │
                     f(t) = -dR/dt    │   f(t) = dF/dt
                     f(t) = h(t)·R(t) │
                                      ▼
          ┌───────────────────────────┴───────────────────────────┐
          ▼                                                       ▼
 ┌─────────────────┐        R(t) = 1 - F(t)              ┌─────────────────┐
 │Reliability: R(t)│ ◄─────────────────────────────────► │Unreliability:F(t│
 └────────┬────────┘                                     └─────────────────┘
          │
          │  h(t) = f(t) / R(t) = - (1/R)·(dR/dt) = -d[ln R(t)]/dt
          ▼
 ┌─────────────────┐
 │ Hazard:  h(t)   │ ──►  R(t) = exp[ - ∫₀ᵗ h(u) du ]
 └─────────────────┘

Reliability Function $R(t)$

The Reliability Function (also known as the survival function) represents the probability that a device or component functions without failure throughout the entire time interval from $0$ to $t$ under specified operating conditions: $R(t) = P(T > t)$ By definition, for any physically realizable system:

• $R(0) = 1$ (the unit is fully operational at time zero).
• $\lim_{t \to \infty} R(t) = 0$ (all physical devices eventually fail).
• $R(t)$ is a monotonically non-increasing function of time ($dR(t)/dt \le 0$).

Cumulative Distribution Function $F(t)$

The Cumulative Distribution Function (CDF), denoted as $F(t)$ or unreliability $Q(t)$, defines the probability that the component fails at or before operating time $t$: $F(t) = P(T \le t) = 1 - R(t)$ Correspondingly, $F(0) = 0$, $\lim_{t \to \infty} F(t) = 1$, and $F(t)$ is monotonically non-decreasing.

Failure Probability Density Function $f(t)$

The Probability Density Function (PDF) $f(t)$ describes the unconditional failure rate per unit time. Mathematically, it is the first derivative of the cumulative failure distribution: $f(t) = \frac{dF(t)}{dt} = -\frac{dR(t)}{dt}$ Integrating the PDF yields the unreliability and reliability over an operating horizon: $F(t) = \int_{0}^{t} f(u)\,du, \qquad R(t) = \int_{t}^{\infty} f(u)\,du$

Hazard Rate / Failure Rate Function $h(t)$

The Hazard Function $h(t)$ (often denoted $\lambda(t)$) represents the instantaneous conditional failure rate. It is the conditional probability that a component fails in the infinitesimal interval $[t, t + \Delta t]$, given that it has survived up to time $t$: $h(t) = \lim_{\Delta t \to 0} \frac{P(t < T \le t + \Delta t \mid T > t)}{\Delta t} = \frac{f(t)}{R(t)}$ Substituting $f(t) = -\frac{dR(t)}{dt}$ into the hazard definition reveals an exact differential relationship: $h(t) = -\frac{1}{R(t)}\frac{dR(t)}{dt} = -\frac{d}{dt}\left[\ln R(t)\right]$ Integrating both sides from $0$ to $t$ with the boundary condition $R(0) = 1$: $\int_{0}^{t} h(u)\,du = -\left[\ln R(t) - \ln R(0)\right] = -\ln R(t)$ Exponentiating both sides yields the fundamental master equation of reliability engineering: $R(t) = \exp\left( -\int_{0}^{t} h(u)\,du \right) = e^{-H(t)}$ where $H(t) = \int_{0}^{t} h(u),du$ is the cumulative hazard function.

Mean Time to Failure (MTTF)

For non-repairable components, the Mean Time to Failure (MTTF) represents the expected value of time-to-failure $T$: $\text{MTTF} = E[T] = \int_{0}^{\infty} t f(t)\,dt$ Applying integration by parts ($u = t$, $dv = f(t)dt = -dR(t)$): $\text{MTTF} = \left[-t R(t)\right]_{0}^{\infty} + \int_{0}^{\infty} R(t)\,dt$ Assuming $\lim_{t \to \infty} t R(t) = 0$ (which holds for all practical engineering distributions), the expected life simplifies to the area under the reliability curve: $\text{MTTF} = \int_{0}^{\infty} R(t)\,dt$

---

2. The Bathtub Curve and Failure Rate Regimes

In physical and industrial equipment, the hazard function $h(t)$ typically varies over the operating lifespan according to the classical Bathtub Curve. The curve comprises three distinct engineering phases:

                         The Classical Bathtub Curve

  Hazard Rate
    h(t) ▲
         │  Phase I: Infant Mortality  Phase II: Useful Life   Phase III: Wear-Out
         │  (Burn-in / DHR)            (Constant / CHR)        (Aging / IHR)
         │
         │ ╲                                                   ╱
         │  ╲                                                 ╱
         │   ╲                                               ╱
         │    ╲                                             ╱
         │     ╲───────────────────────────────────────────╱
         │                 h(t) = λ = constant
         └─────────────────────────────────────────────────────────────►
         0              t_burn-in                       t_wear-out       Time (t)

Phase I: Infant Mortality / Burn-In Period (Decreasing Hazard Rate, DHR)

• Behavior: The hazard rate starts high and decreases rapidly over time ($dh(t)/dt < 0$).
• Physical Root Causes: Substandard manufacturing materials, cold solder joints, microscopic weld porosity, chemical contamination, assembly misalignment, and out-of-tolerance components.
• Engineering Mitigation: Quality engineers implement Environmental Stress Screening (ESS), thermal cycling, power vibration testing, and pre-delivery burn-in procedures. Units that harbor latent flaws fail in the factory before customer shipment. Surviving components emerge into Phase II.
• Statistical Distribution: Commonly modeled using the Weibull distribution with shape parameter $\beta < 1.0$.

Phase II: Useful Life Period (Constant Hazard Rate, CHR)

• Behavior: The hazard rate remains essentially flat and constant over time: $h(t) = \lambda$.
• Physical Root Causes: Failures in this operating window are triggered by purely random, unpredictable external shocks (such as lightning surges, hydraulic pressure spikes, severe foreign object debris, or operator handling error) rather than intrinsic material aging.
• Engineering Significance: Because failure is memoryless during this phase, preventive maintenance (replacing a working component simply because it has accumulated operating hours) provides zero reliability improvement. The part does not age during useful life.
• Statistical Distribution: Strictly governed by the Exponential Distribution ($h(t) = \lambda$, Weibull shape parameter $\beta = 1.0$).

Phase III: Wear-Out Period (Increasing Hazard Rate, IHR)

• Behavior: The hazard rate climbs steeply as operating hours advance ($dh(t)/dt > 0$).
• Physical Root Causes: Cumulative mechanical wear, metal fatigue crack propagation, oxidation, thermal degradation of insulation, bearing friction, seal embrittlement, and chemical depletion of lubricants.
• Engineering Mitigation: Preventive maintenance, scheduled overhauls, and periodic component replacement prior to reaching the wear-out threshold $t_{\text{wear-out}}$ are highly effective.
• Statistical Distribution: Modeled by Weibull distribution with shape parameter $\beta > 1.0$ (e.g., $\beta \approx 2$ to $4$ for mechanical fatigue) or the Normal/Log-normal distribution.

---

3. The Exponential Reliability Model

The exponential distribution is the primary workhorse model for the FE exam because it mathematically embodies the constant hazard rate assumption of Phase II.

Mathematical Formulation

Setting $h(t) = \lambda$ (where $\lambda$ is the constant failure rate expressed in failures per unit time, such as failures per hour): $R(t) = \exp\left( -\int_{0}^{t} \lambda\,du \right) = e^{-\lambda t}$ The cumulative failure probability (unreliability) is: $F(t) = 1 - e^{-\lambda t}$ The probability density function is: $f(t) = -\frac{dR(t)}{dt} = \lambda e^{-\lambda t}$

Mean Time to Failure (MTTF)

Evaluating the mean time to failure for an exponentially distributed component: $\text{MTTF} = \int_{0}^{\infty} R(t)\,dt = \int_{0}^{\infty} e^{-\lambda t}\,dt = \left[ -\frac{1}{\lambda} e^{-\lambda t} \right]_{0}^{\infty} = 0 - \left(-\frac{1}{\lambda}\right) = \frac{1}{\lambda}$ Therefore: $\text{MTTF} = \frac{1}{\lambda} \iff \lambda = \frac{1}{\text{MTTF}}$

The Critical $R(\text{MTTF})$ Relationship

A perennial exam trap concerns the survival probability of a component operating exactly up to its MTTF ($t = \text{MTTF} = 1/\lambda$): $R(\text{MTTF}) = e^{-\lambda (1/\lambda)} = e^{-1} = \frac{1}{2.71828} \approx 0.3679 \quad (36.8\%)$ Uninformed candidates often guess that 50% of components survive to their mean life. In reality, only 36.8% survive to the MTTF under an exponential model, while 63.2% of components will fail prior to reaching the MTTF ($F(\text{MTTF}) = 1 - e^{-1} \approx 0.6321$). The median life $t_{\text{med}}$ (where $R(t) = 0.50$) occurs much earlier: $e^{-\lambda t_{\text{med}}} = 0.50 \implies -\lambda t_{\text{med}} = \ln(0.50) = -0.6931 \implies t_{\text{med}} = \frac{0.6931}{\lambda} = 0.6931 \times \text{MTTF}$

The Memoryless Property

The exponential distribution is the unique continuous distribution possessing the memoryless property: $P(T > t + s \mid T > s) = \frac{P(T > t + s)}{P(T > s)} = \frac{e^{-\lambda(t+s)}}{e^{-\lambda s}} = e^{-\lambda t} = P(T > t)$ In plain physical terms: an old component that has operated for $s$ hours without failure has the exact same probability of surviving the next $t$ hours as a brand-new component straight out of the box. Used equipment does not wear out while operating in the constant failure rate zone.

---

4. MTBF vs. MTTF and System Availability

The FE exam expects clear distinction between repairable and non-repairable terminology.

                 Life Cycle of a Repairable System (MTBF)

  ┌─────────────────────────────── MTBF ───────────────────────────────┐
  │                                                                    │
  ├──────────────────────────── MTTF ──────────────────────────┤─ MTTR ┤
  │                                                            │       │
  ▼                                                            ▼       ▼
┌─────────────────────────────────────────────────────────────┐┌───────┐
│                      Operating State                        ││ Repair│
│                       (System UP)                           ││ (DOWN)│
└─────────────────────────────────────────────────────────────┘└───────┘
0                                                              t_fail  t_repaired

Non-Repairable Items: MTTF

• Used for items discarded upon initial failure (e.g., microprocessors, light bulbs, solid rocket boosters, structural rivets).
• $\text{MTTF} = 1/\lambda$.

Repairable Systems: MTBF and MTTR

• Mean Time Between Failures (MTBF) applies to systems restored to operation through maintenance (e.g., industrial compressors, CNC machine tools, chemical pumps).
• Mean Time to Repair (MTTR) is the average active maintenance time required to diagnose, repair, and test the failed asset.
• The relationship across a recurring operational cycle is: $\text{MTBF} = \text{MTTF} + \text{MTTR}$ In many high-reliability industrial assets where $\text{MTTR} \ll \text{MTTF}$ (e.g., an MTTR of 4 hours vs. an MTTF of 5,000 hours), engineers approximate $\text{MTBF} \approx \text{MTTF}$. On quantitative exam items, however, ensure you include MTTR when specified.

Inherent System Availability ($A_i$)

Availability measures the percentage of operating time an asset is operational: $A = \frac{\text{Uptime}}{\text{Total Time}} = \frac{\text{MTTF}}{\text{MTTF} + \text{MTTR}} = \frac{\text{MTBF} - \text{MTTR}}{\text{MTBF}}$

---

5. Step-by-Step Worked Engineering Calculations

Worked Example 23.1.1: Hazard Rate and Mission Reliability

Problem: A fleet of precision servo actuators used in robotic pick-and-place assembly cells exhibits an exponential failure distribution with an empirical constant failure rate of $\lambda = 0.00025\text{ failures per operating hour}$.

1. Calculate the actuator's Mean Time to Failure (MTTF).
2. Determine the reliability of a servo actuator over a standard industrial operating mission of $t = 1,200\text{ hours}$.
3. What is the probability that an actuator survives an additional $1,200\text{ hours}$, given that it has already operated successfully for $3,000\text{ hours}$ without failure?
4. Calculate the operating time $t_{0.90}$ at which the actuator reliability degrades to 90%.

Solution:

Step 1: Compute MTTF $\text{MTTF} = \frac{1}{\lambda} = \frac{1}{0.00025\text{ hr}^{-1}} = 4,000\text{ operating hours}$

Step 2: Reliability over 1,200 hours $R(1,200) = e^{-\lambda t} = e^{-(0.00025)(1,200)} = e^{-0.30} \approx 0.7408 \quad (74.08\%)$

Step 3: Conditional Survival Probability By the memoryless property of the exponential distribution: $P(T > 3,000 + 1,200 \mid T > 3,000) = P(T > 1,200) = R(1,200) = e^{-0.30} \approx 0.7408$ The prior 3,000 hours of flawless operation neither degrade nor improve the actuator's subsequent 1,200-hour survival probability.

Step 4: Operating Time for 90% Reliability Set $R(t) = 0.90$: $e^{-\lambda t} = 0.90 \implies -\lambda t = \ln(0.90) = -0.10536$ $t = \frac{0.10536}{0.00025} = 421.44\text{ hours}$

---

Worked Example 23.1.2: General Hazard Rate Integration and Wear-Out

Problem: An experimental turbine seal operates in a corrosive chemical environment where its failure rate increases linearly with time according to $h(t) = k t$, where $k = 4.0 \times 10^{-6}\text{ failures/hour}^2$. Find the reliability function $R(t)$ and compute the probability that the seal survives an operating run of $t = 500\text{ hours}$.

Solution:

Step 1: Apply the Master Hazard Integral Equation $R(t) = \exp\left( -\int_{0}^{t} h(u)\,du \right) = \exp\left( -\int_{0}^{t} k u\,du \right) = \exp\left( -\frac{k t^2}{2} \right)$ (Note: This matches a Weibull distribution with shape parameter $\beta = 2$, a Rayleigh distribution typical of linearly accelerating mechanical wear).

Step 2: Evaluate at $t = 500$ hours $-\frac{k t^2}{2} = -\frac{(4.0 \times 10^{-6})(500)^2}{2} = -\frac{(4.0 \times 10^{-6})(250,000)}{2} = -\frac{1.0}{2} = -0.50$ $R(500) = e^{-0.50} \approx 0.6065 \quad (60.65\%)$

---

6. NCEES Reference Handbook Tips & Realistic Exam Traps

• Unit Consistency with Failure Rates: Failure rates are frequently quoted in units of failures per million hours ($10^{-6}/\text{hr}$) or FITs (Failures in Time), where $1\text{ FIT} = 1\text{ failure per } 10^9\text{ hours}$. Always convert $\lambda$ into consistent units matching the mission time $t$ before evaluating $e^{-\lambda t}$.
• $R(\text{MTTF}) \ne 0.50$: Do not fall for exam distractors asserting that 50% of exponential components survive to MTTF. $R(\text{MTTF}) = e^{-1} \approx 0.368$. The median life is $0.693\times\text{MTTF}$.
• Preventive Maintenance on Exponential Components: Questions often describe a system operating in Phase II (useful life) and ask how much reliability increases if preventive replacements are conducted twice as often. The answer is zero increase; because the hazard rate is constant, replacing working components that have not entered wear-out provides no statistical benefit and wastes capital.
• Distinguishing $f(t)$ vs. $h(t)$: The probability density function $f(t)$ is the unconditional failure probability density across the whole population from time zero, whereas $h(t)$ is the conditional failure rate among components that have survived up to time $t$. Because $R(t) \le 1$, $h(t) = f(t)/R(t) \ge f(t)$.