Valeo — Process FMEA for an Assembly Line

An assembly line changed. Its documents hadn't. I rebuilt them

Context
Internship · Valeo Tunisie (Ben Arous)
Role
Process Engineering Intern, with the line's process engineer
Period
Jul–Aug 2025
133failure causes analysed across 30 operations
0 / 13high / medium action-priority risks, every medium one tied to the operator or the environment
111causes stopped inside the stations by automated detection or error-proofing
The line as it runs now: one operator, a four-station rotary table with sensors, cameras, barcode traceability, capacitive tests and screwing control built in, and the end-of-line tester removed

The problem

The end-of-line tester came off a capacitive touch-panel assembly line after months of stable production. Its quality documents went on describing a process that no longer existed.

What I built

  • Genba first: every station of the four-station rotary table, watched and walked through with the line's process engineer, noting where the documents and the line disagreed.
  • A new workstation sheet for the line's single operator, in the site's standard format.
  • A new process flow chart: value-added, non-value-added and control steps, each with its standard symbol.
  • A complete process FMEA on the AIAG-VDA seven-step method: 30 operations, 133 failure causes traced by 6M, each rated for severity, occurrence and detection and given an action priority.

Inside the decisions

01

Where the risk sits

Machines and methods came out clean: all 65 of their failure causes rated low. Every one of the 13 medium risks sits with people (11) or the environment (2).

So the actions went to the operator, not the equipment: an ergonomics review at the manual loading station, and the standards written back into the new workstation sheet.

Failure causes by origin: operator 62 with 11 medium risks, method 43 all low, machine 22 all low, environment 6 with 2 medium; no high-priority risk
02

Why the tester could go

111 of the 133 causes are stopped inside the stations: 65 by automated detection, 46 made impossible or always caught by error-proofing. The other 22 rely on a person's eyes.

Checks built into the process only stay trustworthy if they are themselves checked, so the FMEA adds periodic verification and calibration of the cameras, barcode readers, capacitive testers and screwing control.

What stops each failure cause: 65 by automated machine detection, 46 error-proofed, 22 by human inspection or an alert

The outcome

Three documents that agree with each other and with the line as it runs: no high-priority risk left, and preventive actions aimed at where the remaining risk lives.

Stack

AIAG-VDA FMEA (7 steps)
Action priority (S/O/D)
6M root causes
Process flow chart
Workstation sheet
IATF 16949

What it proves

PFMEAProcess engineeringQualityRisk analysisLean / GenbaQRQCMSAProduction KPIs

Also in Engineering AI