Statistical Process Control (SPC)

The Complete Practical Guide

1) What SPC Is (and What It Is Not)

Statistical Process Control (SPC) is the use of statistical methods to monitor, control, and improve process performance over time. Its purpose is to distinguish normal process variation from special causes and to trigger action before defects are produced.

SPC is not:

  • end-of-line inspection

  • capability reporting only

  • a chart created for audits

  • “plotting data without reacting”

SPC only works if it leads to decisions and actions.

2) Why SPC Exists

Every process varies. SPC helps you answer:

  • Is the process stable?

  • Is the variation predictable?

  • Did something change that requires action?

  • Can the process meet requirements consistently?

Without SPC, problems are found after defects appear.

3) Where SPC Fits in the Quality System

SPC sits downstream of MSA and upstream of capability and control decisions:

Requirements → DFMEA → PFMEA → Control Plan → MSA → SPC → Capability → Reaction / Improvement

If MSA is weak, SPC data is meaningless.

4) Variation — The Foundation You Must Understand

Common Cause Variation

  • Natural, inherent to the process

  • Present all the time

  • Reduced only by changing the process itself

Special Cause Variation

  • Abnormal, unexpected

  • Due to specific events: tool breakage, setup error, wrong material, software bug

  • Must be identified and eliminated quickly

SPC’s primary job is detecting special causes early.

5) Control Charts — The Core SPC Tool

What Control Charts Do

  • Show process behavior over time

  • Separate signal from noise

  • Provide statistically valid decision rules

Control limits are not specification limits.

6) Choosing the Right Control Chart

Variable Data (Measured Values)

  • X̄–R Chart
    For subgroups of moderate size (e.g., 2–10 parts)

  • X̄–S Chart
    For larger subgroups

  • Individuals (I-MR) Chart
    For low-volume production or destructive testing

Attribute Data (Counts or Pass/Fail)

  • p / np Charts — fraction or number of defectives

  • c / u Charts — count of defects per unit

Attribute charts are weaker and should be avoided for critical characteristics when possible.

7) Control Limits vs Specification Limits

This distinction is frequently misunderstood.

  • Specification limits come from design and customer requirements.

  • Control limits come from the process itself.

A process can be:

  • stable but incapable

  • capable but unstable (rare and dangerous)

  • stable and capable (goal)

Never adjust a stable process just because points are within spec.

8) SPC Rules — When to React

Typical out-of-control signals include:

  • a point beyond control limits

  • runs of points on one side of the centerline

  • trends upward or downward

  • abnormal patterns (cycles, sudden shifts)

Each signal must trigger a defined reaction plan.

SPC without reaction rules is decoration.

9) Reaction Plans (The Missing Link)

A reaction plan answers:

  • Who stops the process?

  • What happens to suspect product?

  • How is root cause investigated?

  • When can production restart?

  • How is recurrence prevented?

Without reaction plans, SPC charts are useless.

10) SPC and Control Plans

If SPC is listed in the Control Plan:

  • chart type must be defined

  • subgroup size and frequency must be defined

  • reaction rules must be defined

  • responsibility must be defined

Mismatch between Control Plan and SPC practice is a common audit finding.

11) SPC and Capability (Cpk / Ppk)

Stability Comes First

  • Capability numbers are valid only if the process is stable

  • Unstable process = meaningless Cpk/Ppk

Cpk vs Ppk

  • Cpk: short-term capability (within-subgroup variation)

  • Ppk: long-term performance (overall variation)

Use capability to improve the process, not to argue with customers.

12) SPC and PFMEA

SPC is a detection control in PFMEA.

  • Strong SPC = lower detection rating

  • No SPC or weak reaction = higher detection risk

PFMEA must reflect real SPC effectiveness, not intent.

13) SPC in Low-Volume or High-Mix Environments

SPC is still possible, but adapted:

  • Individuals charts

  • short-run SPC

  • standardized charts with coded data

  • focus on key process parameters instead of product output only

“No volume” is not an excuse to ignore SPC.

14) Common SPC Mistakes

  • using SPC without validated MSA

  • over-adjusting stable processes (tampering)

  • ignoring out-of-control signals

  • confusing control limits with specs

  • using attribute charts when variable data is available

  • collecting data without ownership or action

These mistakes create noise and waste.

15) When SPC Must Be Revisited

Update SPC when:

  • process changes

  • tooling or material changes

  • new suppliers are introduced

  • control limits drift significantly

  • customer complaints or internal failures occur

SPC is dynamic, not static.

16) What Auditors and Customers Expect

They typically check:

  • alignment between PFMEA, Control Plan, and SPC

  • evidence of reaction to out-of-control conditions

  • control limit calculations

  • training of operators and engineers

Most SPC findings are about lack of action, not statistics.

17) Practical Checklist for Effective SPC

  • measurement system is validated

  • chart type matches data

  • control limits are based on data, not assumptions

  • reaction plan exists and is used

  • operators understand when and how to react

  • SPC data is reviewed, not archived

18) SPC vs Inspection (Clear Distinction)

  • Inspection finds defects after they occur

  • SPC prevents defects before they occur

Relying on inspection instead of SPC is expensive and fragile.