A control plan is the document that says, for each step in your process, what is controlled, how, how often, by whom, and what happens when it goes wrong. You build it from your process flow and your PFMEA: every risk the PFMEA says matters should have a control in the plan. A good one is short, specific and used on the floor. A bad one repeats the PFMEA, says “100 percent” everywhere and is never opened after launch.
What a control plan is for
The PFMEA is the thinking about what could go wrong. The control plan is what you actually do about it every day. If the PFMEA says a failed wash would ruin the plating, the control plan says who checks the wash, with what, how often, and what they do if it fails. The Control Plan vs PFMEA guide shows how the two documents connect.
AIAG has published a standalone control plan manual alongside a new edition of APQP, so the layout your customer expects may differ from older forms. Ask which format they want before you build.
What each column is asking
| Column | The question it answers |
|---|---|
| Part or process number and name | Which operation is this, in the same words as the flow diagram? |
| Machine, device or tool | What equipment does the work or does the check? |
| Characteristic (product and process) | What is being controlled: a dimension on the part, or a setting on the process? |
| Special characteristic class | Is this one the customer or the PFMEA says matters most? |
| Specification or tolerance | What is the target and limit, in numbers? |
| Evaluation or measurement method | What gauge or method, and is it capable? |
| Sample size and frequency | How many, how often? |
| Control method | What keeps it in control: error-proofing, a chart, a checklist, an inspection? |
| Reaction plan | What happens when it is out: who stops, what is contained, who is told? |
How to build one
Start from the process flow
Use the same operations, in the same order and with the same names as the flow diagram and the PFMEA. Three documents with three versions of the process is the first thing an auditor sees.
Pull the risks from the PFMEA
Take each failure mode that needs control, starting with the high-severity ones. Every one should end up in the plan, or you should be able to say why it does not need to.
List the characteristics
For each operation, list what you control on the product (a dimension, a finish) and on the process (a temperature, a time, a pressure). Mark the special characteristics.
Write the specification in numbers
The target and tolerance, from the print or the process window. “Within spec” is not a specification.
Choose how it will be measured
The gauge or method, and check it can do the job. A gauge that cannot tell good from bad is not a control. See Gage R&R explained.
Set sample size and frequency from the risk
High severity or poor capability means more checking. A stable, error-proofed step needs less. Do not copy “100 percent” down the page.
Pick a control method that fits the failure
Error-proofing beats a checklist, and a checklist beats “be careful.” Charts suit characteristics that drift. See SPC charts for beginners.
Write a reaction plan someone can follow
Say what the operator does next, who they call, and what happens to parts since the last good check. “Notify supervisor” is not a reaction plan.
A made-up example: the bore operation
| Operation | Characteristic | Spec | Method | Sample | Control | Reaction |
|---|---|---|---|---|---|---|
| 40 Bore on CNC lathe | Bore diameter (special) | 25.000 plus or minus 0.050 mm | Air gauge, calibrated, gauge R&R passed | 5 parts every hour | X-bar and R chart | Stop, contain parts since last good check, call quality |
| 40 Bore on CNC lathe | First-off after setup | All dimensions to print | Setup sheet, CMM | First part each setup | Setup verification | No release until first part passes |
| 40 Bore on CNC lathe | Coolant concentration | 7 to 9 percent | Refractometer | Once per shift | Log and check | Adjust, recheck, tag parts if out |
Mistakes that make a control plan useless
- Copying the PFMEA into the plan. They do different jobs. The plan lists controls, not risks.
- “100 percent inspection” everywhere. It is expensive and not very good at catching defects, because people get tired. Use it only where nothing better exists, and say so.
- Controls that do not match severity. A characteristic that could hurt someone and one that is cosmetic should not get the same sampling.
- Reaction plans that say nothing. If the operator would have to guess, the plan has failed.
- A plan that disagrees with the work instruction. The plan says check every hour and the instruction says every shift. One of them is wrong.
- Never updating it. Every complaint, change or audit finding that changes a control should change the plan. Field Book 1, chapter 5 walks through it.
The Control Plan workbook gives you a layout to build from, and the PFMEA to Control Plan check shows which PFMEA risks have no control yet.
Questions people ask
What is the difference between a control plan and a PFMEA?
A PFMEA identifies and rates risks. A control plan lists the controls that manage them. The PFMEA comes first and feeds the plan.
Do I need a control plan for every part?
Your customer’s requirements decide. In automotive, a control plan is part of PPAP, and most customers expect one for each part or family of similar parts.
What are the control plan phases?
Many customers expect plans for prototype, pre-launch and production, each with the controls that apply at that stage.
How often should a control plan be reviewed?
Whenever the process, tooling, material or a control changes, after any complaint it did not prevent, and on a schedule you set.
Who writes the control plan?
The same cross-functional team that wrote the PFMEA, with the people who run the process involved. They know what can really be checked.