11.3 Switching Rules & Variables Sampling (Z1.9)

11.3 Switching Rules & Variables Sampling (Z1.9)

ANSI/ASQ Z1.4 Dynamic Switching Rules

A critical concept tested on the ASQ CQT examination is that ANSI/ASQ Z1.4 provides its advertised consumer protection only when the dynamic switching rules are rigorously enforced. If an organization operates solely on "Normal" inspection indefinitely without applying switching rules, suppliers who consistently produce poor quality at or worse than the AQL will still experience high acceptance rates over time.

The switching rules dynamically adjust inspection severity between three operational states based on the supplier's recent performance history:

1. Normal Inspection: The standard starting point for all new suppliers and production runs.
2. Tightened Inspection: Applied when quality degrades. Tightened inspection reduces acceptance numbers ($Ac$) while generally maintaining sample size ($n$), creating a much steeper OC curve that protects the consumer by rejecting substandard lots with higher probability.
3. Reduced Inspection: Applied when quality is consistently superior. Reduced inspection cuts the sample size substantially (typically to approximately $40%$ of the normal sample size), lowering appraisal costs while maintaining acceptable producer protection.

ANSI/ASQ Z1.4 SWITCHING RULES STATE DIAGRAM:

                    +-------------------------------+
                    |       REDUCED INSPECTION      |
                    |  (Sample size n ~ 40% normal) |
                    +---------------+---------------+
                                    ^  |
            10 consecutive accepted |  | - 1 lot rejected
           + Table VIII limit met   |  | - Nonconforming between Ac & Re
           + Production steady      |  | - Irregular production
           + QA Authority approval  |  | - Other conditions warrant
                                    |  v
                    +---------------+---------------+
+-----------------> |       NORMAL INSPECTION       |
|                   |      (Default baseline)       |
|                   +---------------+---------------+
|                                   |  ^
|      2 of 5 consecutive rejected  |  | 5 consecutive lots
|      on original inspection       |  | accepted on original
|                                   v  | inspection
|                   +---------------+---------------+
|                   |     TIGHTENED INSPECTION      |
|                   |  (Lower Ac, steep OC curve)   |
|                   +---------------+---------------+
|                                   |
|                                   | 5 lots not accepted in a
|                                   | run of consecutive lots
|                                   | on Tightened inspection
|                                   v
|                   +---------------+---------------+
+------------------ |  DISCONTINUATION OF INSPECTION|
  Supplier corrective| (Shipments halted; vendor QMS |
  action validated  | audit required before restart)|
                    +-------------------------------+

The Mandatory Switching Criteria (ASQ CQT Core Knowledge)

1. Normal to Tightened

Tightened inspection must be instituted immediately when 2 out of 5 (or fewer) consecutive lots have been rejected on original inspection.

• Original Inspection Rule: The 2 rejections must occur on the first presentation of the lots. Re-inspected lots that were sorted or reworked and resubmitted do not count toward this rule.

2. Tightened to Normal

Normal inspection is reinstated when 5 consecutive lots have been accepted on original inspection under tightened inspection.

3. Discontinuation of Inspection

ANSI/ASQ Z1.4-2003 (R2018) states that if the cumulative number of lots not accepted in a sequence of consecutive lots on tightened inspection reaches 5, the acceptance procedures of the standard shall be discontinued. Inspection may not resume under Z1.4 until the supplier has demonstrated improved quality, and it resumes under tightened inspection.

• Operational Consequence: Acceptance of product from that supplier is halted immediately. The supplier cannot ship against the sampling plan until it implements verified corrective actions, eliminates the root causes of variation, and the responsible authority approves the resumption of Z1.4 inspection.

[!CAUTION] The 5-versus-10 Trap: Older textbooks, legacy purchase-order clauses, and study aids built on the cancelled MIL-STD-105E state that inspection is discontinued after 10 consecutive lots remain on tightened inspection. ANSI/ASQ Z1.4-2003 replaced that wording with the stricter 5 lots not accepted criterion, and Z1.4 (not MIL-STD-105E) is the standard in force. If a drawing or contract still invokes MIL-STD-105E by name, follow the contract and document the difference — but answer Z1.4 questions with the 5-lot rule.

4. Normal to Reduced

Transition from normal to reduced inspection is optional but permitted when all four of the following conditions are simultaneously satisfied:

1. The preceding 10 consecutive lots have been accepted under normal inspection.
2. The total number of nonconforming units in the samples from the preceding 10 lots is less than or equal to the applicable limit number specified in ANSI/ASQ Z1.4 Table VIII (or production average is sufficiently stable).
3. Production is in a steady, continuous state (no significant interruptions, machine breakdowns, or extended shutdowns).
4. Reduced inspection is formally approved by the responsible authority (Quality Manager, Chief Metrologist, or Customer QA Representative).

5. Reduced to Normal

Inspection must immediately revert from reduced to normal inspection if any one of the following four conditions occurs:

1. A single lot is rejected ($d \ge Re$).
2. A lot is accepted, but the number of nonconforming units falls between the acceptance number ($Ac$) and rejection number ($Re$) on a reduced plan where $Re > Ac + 1$ (e.g., if $Ac = 1$ and $Re = 3$, observing exactly 2 defective units results in accepting the lot, but immediately forces the inspection scheme back to Normal inspection).
3. Production becomes irregular, intermittent, or experiences an extended pause.
4. Other conditions warrant (e.g., supplier tooling overhaul, critical engineering drawing revision, or customer directive).

---

ANSI/ASQ Z1.9 (MIL-STD-414): Variables Acceptance Sampling

While ANSI/ASQ Z1.4 evaluates discrete attributes (pass/fail, conforming/nonconforming), ANSI/ASQ Z1.9 (the civilian counterpart to military standard MIL-STD-414, harmonized as ISO 3951) governs variables sampling—the inspection of continuous, measurable quantitative characteristics (such as length, diameter, tensile strength, resistance, or coating thickness).

Comparison: Attribute (Z1.4) versus Variables (Z1.9) Sampling

| Operational Parameter | Attribute Sampling (ANSI/ASQ Z1.4) | Variables Sampling (ANSI/ASQ Z1.9) | |---|---|---|| | Data Type | Discrete data (Conforming / Nonconforming, defect counts) | Continuous data (Inches, mm, Rockwell C, PSI, Ohms) | | Sample Size Required | Large (e.g., $n = 50, 80, 125$) | Dramatically Smaller (e.g., $n = 5, 10, 15$) for identical OC protection | | Gaging Required | Fixed limit Go/No-Go gages, functional templates, visual inspection | Calibrated precision instruments (micrometers, CMMs, tensile testers) | | Measurement Speed & Cost | Fast and inexpensive per unit inspected | Slower and higher measurement cost per unit inspected | | Statistical Assumption | Distribution-free (Hypergeometric, Binomial, Poisson) | Strictly requires an approximately Normal (Gaussian) distribution | | Scope of Inspection | Multiple dimensions and features can be checked simultaneously on one gage | Only ONE quality characteristic can be evaluated per sampling plan | | Actionable Quality Information | Binary pass/fail; gives no warning of process drift toward limits | Quantifies sample mean ($\bar{X}$) and variation ($s$), detecting process shift early |

The Trade-off: Sample Size versus Strict Normality

1. Sample Size Advantage: A variables plan in Z1.9 can achieve the exact same statistical risk protection (identical OC curve) as Z1.4 with a 60% to 80% reduction in sample size! For example, where Z1.4 requires inspecting $n = 80$ parts, Z1.9 might require only $n = 10$ parts. This is of monumental economic value when inspection is destructive (e.g., pulling high-cost forged bolts to destruction) or requires lengthy lab cycle times.
2. The Normality Trap: The absolute Achilles' heel of variables sampling is its strict assumption of normality. If the quality characteristic is skewed, bimodal, truncated, or non-normal (such as surface roughness $R_a$, circular runout, or flatness, which are bounded by zero), Z1.9 calculations are completely invalid, leading to severe false acceptance or false rejection errors.

---

Variables Sampling Methods & Quality Index Calculations

Under ANSI/ASQ Z1.9, quality technicians evaluate lot acceptability using either the Standard Deviation Method ($s$-method) or the Range Method ($R$-method) across single or double specification limits.

1. The Standard Deviation Method ($s$-method)

The standard deviation method is the primary and most statistically efficient approach:

1. Draw a random sample of $n$ units and record the exact quantitative measurement ($X_i$) for each.
2. Calculate the sample mean ($\bar{X}$): $\bar{X} = \frac{\sum_{i=1}^n X_i}{n}$
3. Calculate the sample standard deviation ($s$): $s = \sqrt{\frac{\sum_{i=1}^n (X_i - \bar{X})^2}{n - 1}}$

2. Formulating the Quality Index ($Q_U$ and $Q_L$)

The Quality Index expresses the distance between the sample mean and the specification limits, normalized by the sample standard deviation (analogous to a $Z$-score):

• Upper Quality Index ($Q_U$) (distance to Upper Specification Limit, USL): $Q_U = \frac{USL - \bar{X}}{s}$
• Lower Quality Index ($Q_L$) (distance to Lower Specification Limit, LSL): $Q_L = \frac{\bar{X} - LSL}{s}$

QUALITY INDEX GEOMETRY ON NORMAL DISTRIBUTION:

                LSL                Mean X_bar             USL
                 |                     |                   |
                 |              * * * *|* * * *            |
                 |          *          |          *        |
                 |       *             |             *     |
                 |    *                |                *  |
   <-------------+--*------------------+------------------*+------------->
                 |<- Q_L = (X-LSL)/s ->|<- Q_U = (USL-X)/s>|

3. Evaluating Lot Acceptability: Form 1 vs. Form 2

ANSI/ASQ Z1.9 provides two mathematical formats for lot sentencing:

• Form 1 ($k$-method / Acceptability Constant):
  • Used primarily for single specification limits.
  • The quality technician looks up the acceptability constant ($k$) in Z1.9 tables for the designated sample size Code Letter and AQL.
  • Upper limit only: Accept lot if $Q_U \ge k$; Reject if $Q_U < k$.
  • Lower limit only: Accept lot if $Q_L \ge k$; Reject if $Q_L < k$.
• Form 2 ($M$-method / Maximum Allowable Percent Defective):
  • Mandatory for double specification limits (both USL and LSL present).
  • The technician takes $Q_U$ and $Q_L$ and looks up the estimated percent nonconforming ($p_U$ and $p_L$) in Z1.9 Table B-5.
  • Total estimated percent nonconforming is: $p = p_U + p_L$.
  • The technician looks up the Maximum Allowable Percent Defective ($M$) from the Z1.9 table for that AQL.
  • Acceptance Rule: Accept the lot if and only if: $p = p_U + p_L \le M$ (And individually, $p_U \le M$ and $p_L \le M$). Otherwise, reject the lot.

4. The Range Method ($R$-method)

When computational simplicity on the shop floor is paramount, Z1.9 permits the Range Method:

• Rather than computing standard deviation $s$, the inspector calculates sample range $R = X_{\max} - X_{\min}$ (or average range $\bar{R}$ across sub-samples of size 5).
• The Quality Indices become $Q_U = \frac{(USL - \bar{X}) \cdot c}{R}$ and $Q_L = \frac{(\bar{X} - LSL) \cdot c}{R}$, where $c$ is a table factor.
• While simpler to calculate manually without a computer, the range method requires a slightly larger sample size ($n$) to achieve the same statistical protection as the $s$-method.

---

Step-by-Step Worked Numerical Example: Z1.9 Evaluation

Inspection Problem Scenario

An aerospace machine shop inspects a precision cylindrical pin with an outside diameter specified at $0.5000" \pm 0.0020"$ ($LSL = 0.4980"$, $USL = 0.5020"$).

• Inspection Level: General Level II, Normal Inspection, $s$-method, AQL = $1.5%$.
• Lot size: $N = 400$ units.
• From ANSI/ASQ Z1.9 Table A-2, Lot Size 400 yields Code Letter H.
• From Table B-3, Code Letter H under Normal Inspection specifies a sample size of $n = 10$ and maximum allowable percent defective $M = 3.31%$.

Step 1: Collect Sample Data ($n = 10$)

The technician randomly samples 10 pins and records the following diameters (in inches): $0.5008, 0.5012, 0.4998, 0.5004, 0.5015, 0.5002, 0.5010, 0.4995, 0.5006, 0.5010$

Step 2: Compute Sample Mean ($\bar{X}$)

$\sum_{i=1}^{10} X_i = 5.0060"$ $\bar{X} = \frac{5.0060"}{10} = 0.50060"$

Step 3: Compute Sample Standard Deviation ($s$)

$\sum (X_i - \bar{X})^2 = (0.5008 - 0.5006)^2 + \dots + (0.5010 - 0.5006)^2 = 0.00000356$ $s = \sqrt{\frac{0.00000356}{10 - 1}} = \sqrt{\frac{0.00000356}{9}} = \sqrt{0.0000003956} \approx 0.000629"$

Step 4: Calculate Quality Indices ($Q_U$ and $Q_L$)

• Upper Quality Index ($Q_U$): $Q_U = \frac{USL - \bar{X}}{s} = \frac{0.50200" - 0.50060"}{0.000629"} = \frac{0.00140"}{0.000629"} \approx 2.226$
• Lower Quality Index ($Q_L$): $Q_L = \frac{\bar{X} - LSL}{s} = \frac{0.50060" - 0.49800"}{0.000629"} = \frac{0.00260"}{0.000629"} \approx 4.134$

Step 5: Determine Estimated Percent Nonconforming from Table B-5

• For $n = 10$ and $Q_U = 2.23$, Table B-5 yields estimated upper nonconforming: $p_U = 0.52%$.
• For $n = 10$ and $Q_L = 4.13$, Table B-5 yields estimated lower nonconforming: $p_L = 0.00%$ (mean is over 4 standard deviations above LSL).
• Total estimated percent defective: $p = p_U + p_L = 0.52\% + 0.00\% = 0.52\%$

Step 6: Make Lot Acceptability Decision

• Criteria: Compare total estimated $p$ against maximum allowable limit $M = 3.31%$.
• Evaluation: $p = 0.52% \le M = 3.31%$.
• Final Disposition: The lot is ACCEPTED.