A tightening tool passes the MFU with flying colors, Cmk = 1.85 - and yet the customer complains about defective joints from series production. Sounds contradictory? It is not. The Machine Capability Study (MFU) only tests the tool under laboratory conditions. The Process Capability Study (PFU) according to VDI/VDE 2645-3 evaluates the real tightening process - including all disturbing factors from people, material, method, and environment. Anyone who skips the PFU or equates it with the MFU risks invalid process approvals, costly recalls, and weak process capability analysis.
This guide walks you step by step through how to carry out your first standards-compliant PFU in a structured and reliable way, from torque monitoring and torque analysis through to measurement data analysis and documentation.
MFU and PFU: Two studies - two completely different messages
The most common misconception in tightening technology: a passed Cmk capability index proves the quality capability of the process. That is wrong.
Unlike the MFU, the PFU takes into account, in addition to the machine influence, the influence categories of people, material, method, and environment. The MFU answers the question: Is the tool basically capable of doing this? The PFU answers: Does the complete tightening process consistently stay within tolerance under real conditions over time, with sufficient repeatability and process capability?
| Characteristic | MFU - Machine Capability | PFU - Process Capability |
|---|---|---|
| Observation period | Short-term study (hours) | Long-term study (days / weeks) |
| Influencing factors | Only machine & measurement | People, Machine, Material, Method, Environment (5M) |
| Key figures | Cm / Cmk | Cp / Cpk |
| Typical limit | Cmk ≥ 1,67 | Cpk ≥ 1,67 (Automotive) |
| Sample size | at least 50 consecutive measurements | at least 25 subgroups of 5 values each |
| Measurement method (screwdriving technique) | Tool shut-off torque under ideal conditions | Subsequent torque measurement under production conditions |
| Goal | Tool acceptance / calibration release | Process release for serial production |
| Standard basis | VDI/VDE 2645-2 | VDI/VDE 2645-3 |
The MFU is always the first step. A PFU can never be successful if the MFU has previously failed. A capable machine, however, still does not guarantee a capable process. Both studies must be carried out independently and documented separately as part of systematic production monitoring.
What a PFU according to VDI/VDE 2645-3 actually means
The objective of the process capability study for threaded joints is to evaluate and document the quality capability of a tightening process under series-production conditions.
The PFU provides indicators for assessing and continuously improving the tightening process. This includes identifying systematic influences for targeted manufacturing optimization, assessing the effect of corrective actions, evaluating tolerances, and defining the intervention limits of quality control charts.
The guideline uses residual torque measurement for this purpose: after tightening, a calibrated test tool is used to determine the torque required to turn the screw slightly further. The residual torque provides information about settling phenomena and their effect on the preload force. This indirect measurement method enables non-destructive, in-line assembly testing and assembly monitoring as part of series-accompanying assembly quality assurance.
Understanding Cpk and Cmk - and calculating them correctly
The key metrics of the PFU are Cp (process potential) and Cpk (critical process capability index).
Cp and Cpk are statistical indicators used to evaluate a process in production engineering. They show how reliably the targets specified in the specification are achieved. In practice, you use them to calculate Cpk and to derive a meaningful Cmk value as part of your overall process capability analysis.
While the Cp value only indicates the ratio of the specified tolerance to the process variation, the Cpk value additionally takes into account the position of the mean value relative to the center of the tolerance. A process can therefore have a mathematically good spread but still have a poor Cpk - because its mean value is too close to one of the specification limits.
The calculation formula is:
Cpk = min [ (USL - x̄) / (3σ), (x̄ - LSL) / (3σ) ]
Here, USL is the upper and LSL the lower specification limit, x̄ is the process mean, and σ is the standard deviation of the measured values.
A Cpk of 1.67 results in a very low scrap rate of 0.57 ppm (parts per million). In the past, a minimum requirement of Cp and Cpk ≥ 1.33 was common and is still valid in many industries today. In the automotive industry, by contrast, the minimum requirement is Cp and Cpk ≥ 1.67.
Use the interactive calculator to calculate Cpk and your other capability index values directly from your data:
The 6 steps of your first PFU according to VDI/VDE 2645-3
First determine the screw joint class according to VDI/VDE 2862 (A, B, or C) and set the tolerance limits (OSG / USG) for the re-tightening torque. Without defined limits the subsequent Cpk calculation is meaningless. Safety-critical connections of class A require especially tight tolerances.
Before you start with the PFU: Ensure that the screw process is statistically controlled. Perform a pre-run study with a quality control chart over a representative period. If the control chart shows trends, runs or breaches of control limits, the PFU is invalid.
Under real production conditions, collect at least 25 subgroups of 5 measurements each (= 125 values). Important: shift changes, material batch changes and operator changes must be included in the sample. Measure the re-tightening torques directly at finished, tightened connections with a calibrated test instrument.
Check the measurements for normal distribution (e.g., Anderson-Darling test or histogram analysis). The standard formulas for Cp and Cpk apply only to normally distributed data. For non-normal distributions, alternative methods (e.g., percentile method) should be used.
Compute the capability indices: Cpk = min[(OSG - x̄) / (3σ), (x̄ - USG) / (3σ)]. The smaller of the two partial values (Cpo / Cpu) is decisive. In the automotive industry, the lower limit is Cpk ≥ 1.67. If the value is below this, the process is not release-ready and corrective actions to optimize the process should be initiated.
Archive all measurement data, calculation bases and results in a standards-compliant and traceable manner. If PFU is passed (Cpk ≥ required threshold) grant the process release. At the same time, set the intervention limits for quality control charts and define the PFU repeat interval.
Typical mistakes - and how to avoid them right from the start
One particularly critical point in practice: the confidence interval of the Cpk value becomes very large for small samples - which significantly reduces the informative value of your process capability analysis. If you work with 30 instead of 125 measured values, you will indeed obtain a numerical result, but statistically you can hardly trust it.
It is also true that only if the control chart shows no violations of intervention limits or unnatural patterns (trends, runs) over a representative period is the process considered stable and the PFU valid. Stable processes with sufficient repeatability are the foundation of reliable production monitoring and precision measurement technology.
PFU and VDI/VDE 2862: How the standards interact
The PFU according to VDI/VDE 2645-3 draws directly on the tightening case classification of VDI/VDE 2862: the class (A, B, or C) determines how tightly the tolerance limits have to be set - and therefore which Cpk value your process must demonstrate. Safety-critical joints in Class A require the strictest limit values and the most complete documentation. You can find more on tightening case classification in our article VDI/VDE 2862 explained simply: What categories A, B, and C mean for your tightening processes.
The right tool for a standards-compliant PFU
A PFU is only as robust as the measuring equipment used. The decisive factors are:
- Calibrated measuring systems with complete traceability (DAkkS-accredited)
- Residual torque measurement without operator influence for reproducible results and high repeatability
- Automatic data acquisition and statistical evaluation (Cmk, Cpk, histogram, control chart) directly in the system for efficient measurement data analysis
- Standards-compliant archiving of all measurement data for audits and long-term production monitoring
The GWK QUANTEC MCS® is specifically designed for test procedures according to VDI/VDE 2645-3: the reference-free angle measurement with ±1% measurement accuracy between 10 and 100% of the nominal range, the integrated residual torque measurement, and the direct Wi-Fi connection to the analysis software QuanLabPro make the system a reliable tool for series-accompanying PFU and torque monitoring in demanding assembly monitoring environments. All results - tightening curves, Cpk calculations, histograms - are documented in an archivable and audit-proof way, supporting manufacturing optimization and robust assembly quality assurance.
If you want to cover your tool requirements for a PFU campaign flexibly, the GWK ToolRent® rental service offers a cost-effective solution: calibrated, ready-to-use devices are available for weeks or months without investment, ideal for temporary assembly testing or short-term capability studies.
Conclusion: Standards-compliant, documented, audit-proof
The process capability study according to VDI/VDE 2645-3 - often referred to simply as VDI 2645 in practice - is not a one-off mandatory exercise. It is the backbone of your quality assurance in tightening technology and precision measurement technology. Anyone who consistently follows the six steps, avoids typical mistakes, and relies on calibrated measuring equipment establishes robust Cpk and Cmk value evidence that any auditor will accept.
Three core results that every PFU must deliver:
- Proof of process stability (control chart)
- Cpk value ≥ required limit value (for example, 1.67 in automotive applications)
- Complete, standards-compliant documentation of all measurement series
Together with you, we develop the optimal solution for your specific tightening cases - from tool selection and DAkkS-accredited calibration through to complete PFU documentation, including capability index reporting, torque analysis, and integrated process capability for your assembly monitoring.
How often must a PFU be repeated according to VDI/VDE 2645-3?
VDI/VDE 2645-3 does not prescribe a fixed repetition interval. In practice, PFUs are typically repeated after significant process changes (tool change, material change, new operators), after deviations in serial monitoring, or at the customer's request (e.g., for automotive audits). As a guideline, an annual cycle applies or when starting a new part type.
What is the difference between Cpk and Cmk in screw tightening?
Cmk is the critical machine capability index and describes the tool's capability under controlled ideal conditions (briefly, without external disturbances) - typical threshold: Cmk ≥ 1.67. Cpk is the critical process capability index and evaluates the entire screw-tightening process under real production conditions including all factors (operator, material, environment). A high Cmk does not guarantee a high Cpk.
Can a passing Cmk value replace the PFU?
No. The MFU (with Cmk) evaluates only the screw tool under laboratory conditions - it is a necessary prerequisite, but not a replacement for the PFU. Only the PFU according to VDI/VDE 2645-3 under real production conditions demonstrates the actual quality capability of the overall process. Both investigations should be documented in a standards-compliant manner.
Which measurement method does VDI/VDE 2645-3 prescribe for the PFU?
The directive VDI/VDE 2645-3 describes the retorque measurement as the method for the PFU. After tightening, a calibrated inspection tool is applied and the torque required to turn the fastener slightly further is measured. This retorque torque is an indirect measure of the actually achieved preload in the joint.
From what Cpk value is a screw process considered capable?
The required minimum Cpk depends on the industry and the customer. In the automotive industry, Cpk ≥ 1.67 is the standard for safety-critical connections of class A. A Cpk ≥ 1.33 is considered sufficient in many other industries. If Cpk < 1.33, the process is considered not capable and immediate corrective actions must be initiated.
