Process Capability Study for Custom Metal Parts: How Buyers Should Read Cp and Cpk Before Approval
When you source custom metal parts, a process capability study is not just a quality engineer’s spreadsheet exercise. It is one of the clearest ways to judge whether a supplier can repeatedly make parts that fit, assemble, seal, and perform the way your drawing requires. For OEM buyers, sourcing managers, and supplier quality engineers, the real question is simple: can this process hold the dimensions that matter before you release production?
This is where Cp and Cpk come in. These capability indices help you separate suppliers who can quote a tolerance from suppliers who can actually maintain it in production. A capable process reduces line stoppages, sorting costs, late corrective actions, customer returns, and expensive surprises after PPAP or first article approval.
In this guide, we explain process capability study custom metal parts from a buyer-first perspective. You will learn how to read Cp and Cpk correctly, what sample size to ask for, which dimensions should be included, how acceptance thresholds change by risk level, and why casting and machining processes must be judged differently.
Quick Answer
A process capability study uses measured data from a controlled production run to show whether a supplier’s process can consistently produce parts within your specification limits. For custom metal parts, buyers should focus more on Cpk than Cp, because Cpk shows both variation and centering.
| Metric | What It Tells You | Buyer Takeaway |
|---|---|---|
| Cp | Potential capability if the process is centered | Useful, but not enough on its own |
| Cpk | Actual capability considering both spread and centering | Main number to use for approval decisions |
| Pp/Ppk | Overall long-term performance | Helpful for ongoing production review |
As a practical rule:
- Cpk below 1.00: do not approve for production without corrective action.
- Cpk around 1.33: common minimum for many standard critical dimensions.
- Cpk 1.67 or higher: often expected for special, safety, sealing, or tightly mating characteristics.
But no single threshold should be used blindly. You must also check sample size, process stability, gauge reliability, and whether the study covered the right dimensions.
Cp vs Cpk Explained in Plain Buyer Language
Many suppliers send a capability report with both Cp and Cpk, then highlight the larger number. Buyers should do the opposite: start with Cpk.
Cp compares the natural process spread to the engineering tolerance. In simple terms, it answers this question: if the process were perfectly centered, would the spread fit inside the tolerance band?
Cpk goes one step further. It measures how close the process average is to the target and how much variation exists. That makes it more realistic. A process can have a good Cp but still make bad parts if it is drifting toward one spec limit.
| Scenario | Cp | Cpk | What It Means |
|---|---|---|---|
| High Cp, high Cpk | Good | Good | Process variation is small and the process is centered |
| High Cp, low Cpk | Good | Poor | Process could be capable, but it is not centered |
| Low Cp, low Cpk | Poor | Poor | Process spread is too wide and actual output is risky |
| Low Cp, slightly better Cpk | Poor | Poor | Still not capable; do not let presentation hide the issue |
Example: imagine a shaft diameter tolerance of 10.00 +/- 0.05 mm. A supplier’s machining process might show a narrow spread, which gives a decent Cp. But if the machine offsets are set too high and the average diameter is 10.04 mm, the actual margin to the upper limit is small. In that case, Cpk drops, and the process is not safe even though Cp looks respectable.
That is why buyers should ask: is the process capable where it is actually running, not where it could run in theory?
Why Capability Matters for OEM Buyers
Capability studies matter because most sourcing failures do not start with a totally bad supplier. They start with a process that is “almost good enough” during sampling, then slips when volume increases, shifts change, tooling wears, or raw material lots vary.
For OEMs, a poor or incomplete capability review creates downstream cost in places that are much more expensive than machining or casting itself:
- Incoming inspection sorting and containment
- Assembly interference, leaks, vibration, or poor cosmetic fit
- Missed launch dates due to repeated tooling adjustments
- Field returns and warranty claims
- Emergency premium freight and replacement production
- Supplier disputes over whether the drawing or process is at fault
A proper capability study helps you answer five core sourcing questions before approval:
- Can the supplier hold the tolerance repeatedly, not just once?
- Is the process centered near target, or living near a spec edge?
- Which dimensions are risky and need stronger control plans?
- Does the process behave differently by cavity, tool, machine, or shift?
- Should you approve production, approve conditionally, or require improvement first?
Buyer checklist before relying on any capability report:
- Was the study run with production tooling and normal settings?
- Were calibrated gauges used, with acceptable measurement system results?
- Was the study done on actual production material?
- Were all critical-to-quality dimensions included?
- Was the sample sequence time-ordered instead of cherry-picked?
- Did the supplier show both data and conclusions?
How a Process Capability Study for Custom Metal Parts Should Be Run
A credible process capability study for custom metal parts should come from a stable production run, not from hand-selected parts measured after repeated machine tweaking. Buyers should expect the supplier to produce parts under normal manufacturing conditions, then measure them in order and analyze the results against drawing specifications.
At minimum, the study package should include:
- Part number, revision, date, machine, tool, cavity, and operator or shift details
- Measurement method and gauge identification
- Specification limits and target values
- Raw measurement data for each sample
- Histogram or distribution plot
- Control chart or stability evidence when relevant
- Cp/Cpk calculations by characteristic
- Clear disposition for any dimensions below target threshold
If the report only gives a summary line like “Cpk = 1.45, approved,” that is not enough. Buyers should ask to see raw data and the dimensions selected for study.
How Many Samples Should Buyers Ask For?
Sample size is one of the most common weak points in supplier submissions. A study based on 5 or 10 parts may look clean simply because it is too small to show real variation. For production approval, buyers need enough data to see actual process behavior.
| Situation | Typical Sample Expectation | Buyer Comment |
|---|---|---|
| Initial feasibility or prototype check | 10-20 parts | Useful for early risk review, not full production approval |
| PPAP or first production capability study | 30 consecutive parts minimum | Common baseline for calculating Cp/Cpk |
| Multi-cavity tooling | 30 per cavity if critical features differ by cavity | Do not average away cavity problems |
| High-risk or safety-critical dimensions | 50-125+ parts depending on requirement | More evidence is justified for serious failure risk |
Thirty consecutive parts is a widely used starting point because it gives a more meaningful picture of process spread. But buyers should not treat “30” as magic. If the process has multiple cavities, multiple spindles, multiple fixtures, or large thermal drift over time, one simple 30-piece run may hide the real problem.
Ask how the parts were sampled:
- Consecutive pieces from one setup?
- Across multiple hours?
- Across cavities or machines?
- Before and after tool wear adjustment?
The more variation sources your program faces in normal production, the more your study should reflect them.
Which Dimensions Should Be Included in the Study?
Not every drawing dimension needs a full capability study. Buyers should focus on the dimensions that affect function, assembly, safety, appearance, sealing, or downstream process fit.
In practice, the best study dimensions usually fall into these groups:
| Dimension Type | Examples | Why It Matters |
|---|---|---|
| Fit and assembly features | Hole position, boss diameter, slot width | Affects mating and interchangeability |
| Sealing surfaces | Flatness, groove width, sealing diameter | Leak risk rises quickly with variation |
| Critical datums and relationships | Parallelism, concentricity, position | Drives overall geometric function |
| Thin-wall or distortion-prone features | Wall thickness, rib height, warp-sensitive areas | High sensitivity to process instability |
| Customer-designated special characteristics | SC, CC, KC, or safety symbols on drawing | Requires stronger control and evidence |
Buyers should avoid a common trap: letting the supplier choose only easy dimensions. If a capability package contains generous overall lengths but ignores positional tolerance, sealing diameter, or casting wall thickness near a critical feature, it is incomplete.
Dimension selection checklist for buyers:
- Features that drive assembly success
- Features with the tightest tolerances
- Features historically linked to defects
- Features affected by tool wear or thermal change
- Geometric tolerances, not just simple sizes
- Any drawing-marked critical characteristics
Acceptance Thresholds: What Cp and Cpk Should Buyers Require?
There is no universal threshold that fits every custom metal program, but there are practical acceptance bands buyers can use. The right threshold depends on part function, process maturity, inspection strategy, and failure cost.
| Cpk Range | Interpretation | Typical Buyer Action |
|---|---|---|
| < 1.00 | Not capable | Reject for routine production approval; require corrective action |
| 1.00-1.32 | Marginal | Conditional approval only if risk is low and containment exists |
| 1.33-1.66 | Generally capable | Common acceptance range for standard critical dimensions |
| >= 1.67 | Strong capability | Preferred for high-risk, safety, sealing, or difficult-to-inspect features |
Useful buyer rule: require higher Cpk where failure cost is high and detection is hard. For example, a hidden internal sealing bore should have a stronger capability expectation than a non-functional outside profile with easy 100% inspection.
However, capability numbers only matter if the process is stable. A nice Cpk from an unstable process is false comfort. Ask whether the supplier verified statistical control before calculating capability. If the control chart shows instability, the Cpk result is not reliable.
Casting vs Machining: Why Capability Risks Are Different
Buyers sourcing custom metal parts often work with both casting and machining suppliers, or with suppliers that combine the two. Capability interpretation should change based on the process, because the variation mechanisms are different.
Capability Issues in Casting
Casting processes such as die casting, investment casting, or sand casting usually face more inherent variation in shrinkage, porosity influence, mold wear, parting line movement, fill behavior, and thermal distortion. Near-net shape capability can be good on some features, but weaker on others, especially across complex geometry.
Common casting-related capability concerns include:
- Dimensional shift by cavity or mold zone
- Flatness and warpage after cooling
- Wall thickness inconsistency in thin sections
- Datum instability caused by as-cast surfaces
- Feature movement after trimming, heat treatment, or shot blasting
For cast parts, buyers should be careful not to assume that one good machined feature proves the whole part is capable. The casting base process and datum strategy still matter.
Capability Issues in Machining
Machining processes such as CNC turning, milling, drilling, and grinding often deliver better repeatability on well-fixtured dimensions, but they are not automatically safe. Capability can deteriorate through tool wear, poor chip evacuation, fixture variation, machine warm-up effects, burr influence, and inconsistent workholding.
Common machining-related capability concerns include:
- Drift over time as tools wear
- Size variation after offset changes
- Position errors from refixturing
- Roundness or cylindricity issues hidden by simple size checks
- Differences between machines in a shared production cell
| Process | Main Variation Drivers | Buyer Focus |
|---|---|---|
| Casting | Shrinkage, mold condition, cooling, distortion | Datum stability, wall variation, cavity differences, post-process movement |
| Machining | Tool wear, fixturing, machine offsets, thermal drift | Time-based drift, setup repeatability, multi-machine consistency |
| Cast + machined part | Both of the above | How casting variation affects machined reference surfaces and final GD&T |
For many custom components, the real sourcing challenge is not casting capability or machining capability in isolation. It is the interaction between them. A cast datum that shifts slightly can force machining to chase the condition, reducing true process capability on the final part.
Red Flags Buyers Should Never Ignore
Some capability reports look formal but still hide risk. These are the warning signs that should trigger deeper review before supplier approval.
- Only Cp is shown, not Cpk. This often hides poor centering.
- Sample size is too small. Ten pieces is not a strong production capability study.
- No raw data is included. You cannot verify stability or outliers.
- “Best of” parts were selected. Capability must reflect real production output.
- One combined study covers multiple cavities. Cavity-specific defects disappear in averaging.
- Measurement system was not validated. Bad gauge data can create fake capability.
- The report excludes the tightest or most functional dimensions. Easy dimensions do not protect your assembly.
- Cpk is acceptable but the mean is too close to a spec limit. The process may be one tool adjustment away from failure.
- No control plan exists for marginal characteristics. Capability without an action plan is incomplete.
If you see two or more of these red flags, consider requiring one of the following before approval:
- New capability run under monitored conditions
- Cavity-by-cavity or machine-by-machine breakdown
- Corrective action with evidence of centering improvement
- Tighter process controls plus temporary 100% inspection
- Design tolerance review if the requirement is unrealistic for the chosen process
How Buyers Should Use Capability Results in Approval Decisions
A process capability study should support a sourcing decision, not replace judgment. Buyers should combine Cp/Cpk with drawing review, PFMEA, control plan, gauge studies, first article results, and supplier process knowledge.
A practical approval framework looks like this:
| Study Result | Risk Level | Recommended Buyer Response |
|---|---|---|
| All critical dimensions meet target Cpk with stable data | Low | Approve production and monitor through launch |
| Most dimensions pass, one or two are marginal but controllable | Medium | Conditional approval with containment and improvement timeline |
| Critical characteristic below threshold | High | Do not fully approve; require process correction first |
| Data quality is weak or study setup is questionable | Unknown | Request new study before making final sourcing commitment |
Conditional approval can make sense if the supplier has a clear improvement plan, the feature is inspectable, and your launch timing requires controlled risk. But buyers should avoid normalizing weak capability. Temporary containment has a habit of becoming permanent cost.
FAQ
Is Cp or Cpk more important for custom metal parts?
Cpk is usually more important because it reflects actual process performance, including centering. Cp only shows theoretical potential if the process is centered.
What is a good Cpk for OEM metal part approval?
For many standard critical dimensions, 1.33 is a common minimum. For high-risk, safety, or sealing features, buyers often expect 1.67 or higher. The right threshold depends on function and failure cost.
Can a supplier pass with a low Cp but acceptable parts?
Possibly for a short run, but it is not a strong production signal. Low Cp means the natural variation is too wide for the tolerance. The process may only be passing because of luck, sorting, or constant adjustment.
How many parts are needed for a capability study?
Thirty consecutive parts is a common minimum for an initial production capability study. More may be needed for multi-cavity tools, multiple machines, long-cycle variation, or critical dimensions.
Should cast parts and machined parts use the same capability target?
Not automatically. Casting and machining have different variation mechanisms. The acceptance target may be similar for the final critical characteristic, but the buyer should judge feasibility, process controls, and improvement path differently.
What if the process is not capable but the drawing tolerance is unrealistic?
Then the right next step is not blind approval. Buyers should review function, tolerance stack-up, process selection, and cost tradeoffs with engineering and the supplier. Sometimes the process must improve; sometimes the drawing should change.
Internal Links for Buyers Researching Custom Metal Part Quality
If you are evaluating suppliers for production approval, these related topics are worth reviewing alongside capability studies:
- Custom CNC machining services for understanding repeatability, fixture strategy, and tolerance control
- Aluminum die casting services for learning how casting variation affects final dimensional outcomes
- Quality control for custom metal parts for inspection planning, gauge control, and defect prevention
- PPAP for custom manufactured parts for integrating capability study results into formal approval workflows
Final Takeaway
For OEM buyers, the purpose of a process capability study is not to collect another document. It is to reduce sourcing risk before that risk turns into scrap, delays, line downtime, or customer complaints. If you remember one rule, let it be this: do not approve a custom metal part process based on Cp alone, summary numbers alone, or easy dimensions alone.
Ask for a real production study, focus on Cpk, review the dimensions that matter most, and judge the result in the context of the actual manufacturing process. Casting, machining, and hybrid processes all create capability differently. Good buyers know the difference, and they approve suppliers based on evidence, not optimistic formatting.
That is how a process capability study custom metal parts becomes useful: not as a statistical formality, but as a practical sourcing tool for better approvals and fewer production surprises.