2.3 Applied Engineering Statistics & Quality Control Probability

2.3 Applied Engineering Statistics & Quality Control Probability

Applied Statistics in Mechanical Engineering

Engineering statistics and quality control provide quantitative tools to manage manufacturing variability, assess component reliability, and guarantee product compliance with tolerance specifications. In modern mechanical production—from CNC machining and casting to automotive assembly—no two manufactured parts are identical. The PRC Mechanical Engineering Licensure Examination (MELE) tests candidates on descriptive sample statistics, continuous/discrete probability distributions, confidence interval estimations, simple linear regression, and statistical process control (SPC) control charts ($\bar{X}$ and $R$ charts).

---

1. Descriptive Statistics & Measure of Dispersion

A set of sample measurements $x_1, x_2, \dots, x_n$ is summarized by measures of central location and dispersion:

• Sample Mean (Arithmetic Average): $\bar{x} = \frac{1}{n} \sum_{i=1}^{n} x_i$
• Sample Variance: $s^2 = \frac{1}{n-1} \sum_{i=1}^{n} (x_i - \bar{x})^2$ (Note: Division by $n-1$ provides an unbiased estimate of population variance $\sigma^2$.)
• Sample Standard Deviation: $s = \sqrt{s^2}$
• Coefficient of Variation (CV): $\text{CV} = \frac{s}{\bar{x}} \times 100\%$

---

2. Probability Distributions in Engineering

Continuous Distributions: Normal (Gaussian) Distribution

The normal distribution models continuous dimensional tolerances, material tensile strengths, and fatigue lives. Standard Normal Variable ($Z$): $Z = \frac{X - \mu}{\sigma}$ where $\mu$ is population mean and $\sigma$ is population standard deviation.

Empirical Rule (68-95-99.7 Rule):

• $\mu \pm 1\sigma$ contains 68.27% of all observations.
• $\mu \pm 2\sigma$ contains 95.45% of all observations.
• $\mu \pm 3\sigma$ contains 99.73% of all observations.

Worked Numerical Example: Tensile Strength Defect Rate

The tensile strength of an aluminum-copper alloy shaft batch is normally distributed with mean $\mu = 450 \text{ MPa}$ and standard deviation $\sigma = 15 \text{ MPa}$. Engineering design specifies a minimum acceptable yield strength of $420 \text{ MPa}$. What percentage of manufactured shafts will fail to meet specification?

Step 1: Calculate $Z$-score for $X = 420 \text{ MPa}$ $Z = \frac{X - \mu}{\sigma} = \frac{420 - 450}{15} = \frac{-30}{15} = -2.00$

Step 2: Look up Cumulative Probability $P(Z < -2.00)$ Using standard normal distribution cumulative tables: $P(Z < -2.00) = 0.022750 \approx 0.0228 \text{ or } 2.28\%$

Thus, approximately $2.28%$ of the shafts will fall below minimum tensile specification.

---

Discrete Distributions: Binomial & Poisson

---

3. Confidence Intervals & Linear Regression

Confidence Interval for Population Mean $\mu$:

When sample size $n \ge 30$ (or population $\sigma$ is known): $\text{CI} = \bar{x} \pm Z_{\alpha/2} \left(\frac{\sigma}{\sqrt{n}}\right)$ For $95%$ confidence ($\alpha = 0.05$), $Z_{0.025} = 1.960$. For $99%$ confidence, $Z_{0.005} = 2.576$.

Simple Linear Regression:

Models relationship between independent variable $x$ (e.g., machining speed) and dependent variable $y$ (e.g., surface roughness): $\hat{y} = \beta_0 + \beta_1 x$

• Slope coefficient: $\beta_1 = \frac{n\sum xy - (\sum x)(\sum y)}{n\sum x^2 - (\sum x)^2}$
• Intercept: $\beta_0 = \bar{y} - \beta_1 \bar{x}$

---

4. Statistical Quality Control (SQC) Charts ($\bar{X}$ and $R$ Charts)

Statistical Process Control monitors manufacturing processes using periodic sub-group samples of size $n$ drawn across $k$ production runs.

• Subgroup Mean: $\bar{X}i = \frac{1}{n} \sum{j=1}^n x_{ij}$
• Subgroup Range: $R_i = x_{\text{max}} - x_{\text{min}}$
• Overall Process Mean: $\bar{\bar{X}} = \frac{1}{k} \sum_{i=1}^k \bar{X}_i$
• Average Range: $\bar{R} = \frac{1}{k} \sum_{i=1}^k R_i$

Standard 3-Sigma Control Limit Formulas:

Mean Chart ($\bar{X}$ Chart): $\text{UCL}_{\bar{X}} = \bar{\bar{X}} + A_2 \bar{R}$ $\text{LCL}_{\bar{X}} = \bar{\bar{X}} - A_2 \bar{R}$

Range Chart ($R$ Chart): $\text{UCL}_R = D_4 \bar{R}$ $\text{LCL}_R = D_3 \bar{R}$

(The factors $A_2, D_3, D_4$ depend strictly on subgroup size $n$ derived from statistical tables.)

---

Worked Numerical Example: Bearing Sleeve Quality Control Chart

An automated lathe produces precision bearing sleeves. Subgroups of $n = 5$ sleeves are sampled across $k = 20$ inspection intervals. Statistical summary yields:

• Overall grand mean $\bar{\bar{X}} = 50.020 \text{ mm}$
• Average range $\bar{R} = 0.080 \text{ mm}$

From standard SPC tables for subgroup size $n = 5$: $A_2 = 0.577, \quad D_3 = 0, \quad D_4 = 2.114$

Calculate the Upper Control Limit (UCL) and Lower Control Limit (LCL) for both the $\bar{X}$ chart and $R$ chart.

Step 1: Compute Control Limits for $\bar{X}$ Chart $\text{UCL}_{\bar{X}} = \bar{\bar{X}} + A_2 \bar{R} = 50.020 + (0.577 \times 0.080) = 50.020 + 0.04616 = 50.06616 \approx 50.066 \text{ mm}$ $\text{LCL}_{\bar{X}} = \bar{\bar{X}} - A_2 \bar{R} = 50.020 - (0.577 \times 0.080) = 50.020 - 0.04616 = 49.97384 \approx 49.974 \text{ mm}$

Step 2: Compute Control Limits for $R$ Chart $\text{UCL}_R = D_4 \bar{R} = 2.114 \times 0.080 = 0.16912 \approx 0.169 \text{ mm}$ $\text{LCL}_R = D_3 \bar{R} = 0 \times 0.080 = 0.000 \text{ mm}$

Thus, the $\bar{X}$ chart limits are $49.974 \text{ mm}$ to $50.066 \text{ mm}$, and the $R$ chart limits are $0.000 \text{ mm}$ to $0.169 \text{ mm}$.

---

5. Six Sigma & Process Capability ($C_p, C_{pk}$)

Process capability measures the ability of a process to produce parts within engineering specifications ($LSL$ to $USL$).

• Process Capability Index ($C_p$) (Potential capability assuming centered process): $C_p = \frac{\text{USL} - \text{LSL}}{6\sigma}$
• Process Capability Ratio ($C_{pk}$) (Actual capability accounting for mean off-centering): $C_{pk} = \min\left( \frac{\text{USL} - \mu}{3\sigma}, \frac{\mu - \text{LSL}}{3\sigma} \right)$

Process Capability Interpretation:

• $C_p < 1.0$: Process is Incapable; produces non-conforming defective parts.
• $C_p = 1.0$: Process is Barely Capable (3-Sigma quality, ~2,700 defects per million).
• $C_p \ge 1.33$: Process is Capable (4-Sigma quality standard).
• $C_p \ge 2.0$: Process achieves Six Sigma Quality, yielding no more than 3.4 Defects Per Million Opportunities (DPMO) (with a standard $1.5\sigma$ mean shift, $C_{pk} = 1.50$).

---

MELE Exam Tips & Shortcuts

• Sample vs Population Variance: If the problem asks for sample variance from $n$ measurements, make sure your scientific calculator is in Sample mode ($s$ or $s_{n-1}$), not Population mode ($\sigma$ or $\sigma_n$).
• $C_p$ vs $C_{pk}$ Relationship: $C_{pk} \le C_p$ always. $C_{pk} = C_p$ only when the process mean $\mu$ is perfectly centered at the midpoint $\frac{\text{USL}+\text{LSL}}{2}$.
• Control Limits vs Specification Limits: Control limits ($\text{UCL}, \text{LCL}$) are computed from natural process variation ($3\sigma_{\bar{x}}$). Specification limits ($\text{USL}, \text{LSL}$) are set by customer design requirements. NEVER confuse $A_2\bar{R}$ control limits with design tolerances!