邱培其 | Draft localized on 2026-05-21

The 2026 AIAG-VDA SPC draft matters because it does more than refresh terminology. It changes how manufacturers should think about process capability, process stability, control chart selection, and evidence for IATF 16949 customer audits. For companies that rely on SPC dashboards, Cpk/Ppk reports, OCAP workflows, or supplier quality scorecards, these changes may affect both software configuration and internal quality procedures.

MiDFUN follows these changes closely because MiDFUN SPC Statistical Process Control Software supports real-time control charts, process capability analysis, automated instrument data collection, and quality system integration for manufacturing sites that need audit-ready SPC records.

Quick Summary of the Six Major Changes

1. Cpk and Ppk: Stability Becomes the Key Difference

In many legacy SPC practices, Cpk is treated as a short-term capability index while Ppk is treated as an overall performance index. The 2026 draft reframes that logic. The practical question becomes: has the process been shown to be statistically stable? If stability is not investigated or not satisfied, manufacturers should be careful about calling the result process capability.

This is important for audit conversations. A high index value without evidence of stability can create false confidence. Under the draft logic, control chart evidence and stability rules become part of the capability story, not a separate appendix.

2. Cw and Cwk Bring Back Within-Subgroup Insight

The draft introduces Cw and Cwk to represent within-subgroup capability. This matters because engineers still need a way to separate short-term variation from longer-term drift, tool wear, lot-to-lot differences, environmental cycles, or setup changes.

A useful interpretation pattern is simple: if Cwk is much better than Ppk, the process may have a strong short-term potential but suffers from long-term shifts. That difference should trigger investigation into special causes, not just a cosmetic report adjustment.

3. Capability Thresholds May Become More Sample-Aware

Manufacturers are used to values such as 1.33 and 1.67. The draft points toward a more statistical interpretation: a smaller sample gives less certainty, so the required observed index may need to be higher to support the same level of confidence. For launch quality, APQP, PPAP, and customer submission packages, this means sample size documentation becomes even more important.

4. Analysis Charts and SPC Charts Serve Different Jobs

The draft distinguishes between retrospective analysis and forward-looking control. An analysis chart is used to study historical process behavior and determine whether the process is stable enough to establish limits. An SPC chart applies that knowledge to monitor future production in real time.

This distinction prevents a common mistake: treating calculated limits from unstable historical data as if they were valid real-time control limits. Before a plant locks limits into an SPC system, it should review distribution behavior, special causes, subgrouping strategy, and chart type selection.

5. Stability Rules Need to Be Explicit

Out-of-control detection should not live only in a spreadsheet or operator habit. Rules for points beyond limits, runs, trends, cycles, and other unusual patterns should be configured in the SPC system, linked to response workflows, and reviewed during audits. This is where software design matters: rules, alerts, OCAP actions, and records must be traceable.

6. OC Curves and ARL Make Detection Power Visible

Different control charts detect different kinds of process shifts. OC curves and ARL help engineers understand how quickly a chart is expected to detect a shift and how often false alarms may occur. This is especially relevant for high-volume manufacturing, semiconductor processes, automotive safety characteristics, and operations where small sustained shifts can become expensive.

What Manufacturers Should Do Now

• Review current SPC reports and identify where Cpk, Ppk, and within-subgroup variation are used.
• Confirm whether process stability evidence is documented before capability conclusions are made.
• Check whether control chart limits are based on stable historical analysis rather than raw historical data.
• Verify that out-of-control rules are configured consistently across plants, product lines, and customer programs.
• Prepare for final manual review once the official Red Volume is released.

References: AIAG & VDA SPC Manual presale, AIAG SPC 2nd Edition notice, and the original MiDFUN Chinese article.