A capability index can look reassuring while concealing an unstable process or a poorly chosen dataset. Before using Cpk to describe a casting dimension, confirm how the data were collected, whether the process is stable and whether the statistical assumptions fit the characteristic.

The index is a summary of a defined process under defined conditions. It is not a permanent property of the casting design, and it should not be used as a substitute for the actual acceptance requirements on delivered parts.

Identify the process represented by the data

Record the manufacturing route, measurement method, time period and relevant grouping. Combining different cavities, treatment routes or setup conditions can produce a distribution that does not represent one consistent process. A single index may hide the differences between those groups.

The inspection method must also be suitable. If measurement variation is a significant contributor, the observed spread may not describe manufacturing variation alone. Review the measurement system before drawing a strong capability conclusion.

Examine stability before comparing with limits

NIST describes capability as a comparison between process performance and specification limits. That comparison is meaningful only when its underlying conditions are understood. A process drifting over time can produce an attractive summary over a short interval and an unsatisfactory result later.

Review the data in sequence rather than only as a histogram. A trend or step change may be easier to see in time order. Do not remove inconvenient results without a documented technical reason and an explanation of how the remaining dataset is interpreted.

Do not force every characteristic into a normal model

Some measured characteristics have distributions that do not suit a simple normal assumption. The statistical method should follow the data and the physical process. Transformations or alternative models require justification and should be explained in the report.

A spreadsheet that automatically calculates Cpk does not establish that its assumptions are appropriate. The responsible analyst should document the method, sample basis and limitations. Avoid presenting a point estimate as exact knowledge of future performance.

Keep capability separate from lot acceptance

A capable process can still produce a nonconforming part, and a small accepted lot does not by itself prove long-term capability. The shipment’s inspection and acceptance plan remains applicable. Capability analysis helps understand and improve the process; it is not a blanket exemption from specified checks.

For illustration, a dimension may be centered during one machining setup but shift after a fixture change. Reporting the earlier index without identifying the change would misrepresent the current route. Review whether new evidence is needed when the process changes materially.

Agree the reporting requirement early

When requesting capability evidence from TripleAce Castings, identify the characteristics, required study basis and applicable customer criteria. This allows production and measurement data to be collected deliberately. Reconstructing a study from unrelated inspection sheets after shipment often produces a weaker result.

Statistical sources

NIST: Assessing Process Capability and NIST: What Is Process Capability? explain the statistical framework. This article does not prescribe a universal Cpk threshold for casting acceptance.