Sumitomo Electric Wiring Systems

Quality Leadership

Quality is built at the station, not inspected at the end. Each topic follows the same PDCA loop, backed by line data.

PPlanDDoCCheckAAct

PDCA · Defect ratio ownership on Toyota and PSA harnesses assembly lines

My Quality Leadership Approach

Zero defects is a team habit

8–9

zero-defect teams per week

0.07%

best assembly line DR, WK45

0.09%

assembly lines DR, WK45

SEWS quality ranking, 61 sites

PPlan

Defect-ratio target per line and per team; every defect logged by team, station and cause.

DDo

Daily defect discussion on the line; abnormality forms; stop and escalate when equipment lets defects pass.

CCheck

Weekly defects-per-team trend and worst line / worst defect review.

AAct

Countermeasures become jigs, training and standards; best practice shared across teams and sites.

Assembly production lines defect ratio (%)

Target 0.16%ActualOut of target

Across about 40,000 harnesses a week, the assembly lines stay near 0.1% defect ratio; only WK35 and WK39 went above the 0.16% target.

PDCA · Pipe line defect reduction · Green Belt DMAIC · July 2022

Root Cause Analysis

From 178 defects to their root causes

178

defects in the July baseline

13 → 1

damage component, team A

4.93 → 5.22σ

sigma level, baseline → target

2.1% → 1.5%

defect ratio, baseline → target

PPlan

July baseline: 8,379 harnesses, 178 defects, yield 99.987%, 4.93σ, OEE 91%. Pareto per team (A, B, C).

DDo

Fishbone and 5 Why on the top defects: harness not fixed to the stand at insertion; jig not guiding the bender operator.

CCheck

Damage component on team A fell from 13 to 1 after the fix.

AAct

Insertion method trained, conveyor edges covered with rubber, jig marks added; targets set at 1.5% DR, 5.22σ, 91.6% OEE.

Pareto: Pipe line defects, before vs after

Before (July)After

Wrong dimension and damage component led the Pareto, so the team attacked them first. After the countermeasures: 145 → 58 defects across the 8 types.

Problem	Root cause	Countermeasure
Damage component & scratched pipe	Harness not attached to the stand correctly at insertion (station 1)	Method training at insertion; rubber covers on conveyor edges
Wrong component orientation	The jig does not guide the bender operator	Position mark on the jig; jig redesign

PDCA · Turning every fix into a standard the next shift follows

SOPs & Work Instructions

If it is not written, it is not standard

Hourly

checks on critical stations

change control on every jig

fishbone in each control plan

Shadow

training before new operators work alone

PPlan

Control plan per line: 6M fishbone, root causes, last-5-days target and a repair plan.

DDo

Train the right method at the station; new operators work as shadows first; hourly checks on critical stations.

CCheck

Lean audit point S4: checklists, PFUs and MOMs defined and used; abnormality forms as evidence.

AAct

4M changes (machine, jig, method, material) only through engineering; standards updated after every countermeasure.

01Abnormality found

02Stop & contain

03Report with abnormality form

04Root cause (6M)

05Update SOP / WI

06Train & verify

Wrong fork after a 4M changeNarrow fork installed by mistake; the correct wide fork restored and the change routed through engineering.

New team on Roof lineUnplanned dimension defects; the new team moved to shadow training before working alone.

Test board passing terminal back-outBoard stopped and reported to engineering; abnormality form raised as evidence.

PDCA · Jigs and vision that make the wrong way impossible

Poka-Yoke: Mistake-Proofing

Design the mistake out, don't inspect it out

mistake-proofing devices

jigs redesigned / modified in one quarter

connector codes covered by AI vision

PPlan

Pick defects caused by method, not by people: clip, branch length, wrong clip, bent pins.

DDo

Redesigned clip jig, restored the correct fork, added a branch-length gauge, started a jig that rejects the wrong clip, AI vision for bent pins.

CCheck

Defects tracked per team after each device; vision images reviewed at each stage.

AAct

Devices standardized on the line; ideas shared through Yokoten.

1Clip jig redesignOld jig trapped the clip tail, so operators skipped it. The new shape gives free access.

2Correct forkA narrow fork had replaced the right one; the wide fork is back under 4M control.

3Branch-length gaugeA gauge on the clip rosette fixes the branch length on PDB LHD.

4Wrong-clip-proof jigIn design: the wrong clip, used 4× in the harness, will not fit the one position meant for the right clip.

5AI bent-pin detectionVision finds connectors 521/522 even when shifted, and boxes any bent pin.

PDCA · DOWNTIME, mapped to real actions on the line

Eliminating the 8 Wastes

Every waste had an owner and an action

8 / 8

wastes with an action

+5%

efficiency, assembly line

≈€23K

material saved per year

4 → 3

pre-blocks on Roof B618

PPlan

Walk the line with the 8 wastes in mind; quantify each with the cycle-time, defect and material data.

DDo

Act on each waste: line balancing, station merges, kitting cleanup, material trim.

CCheck

Check efficiency, defect ratio and material use week over week.

AAct

Keep what worked in the build plan and the lean audit.

Waste	What we found	What we did
Defects	Pipe line baseline 2.1% DR	RCA + poka-yoke; damage component 13 → 1
Overproduction	Pack drifting from plan	Hourly plan vs pack board
Waiting	Inverter 2 waits 30 s per cycle	Line rebalancing with time studies
Non-utilized talent	Uneven operator effectiveness	Kaizen to raise every operator's effectiveness
Transportation	Kitter and tape-back as two stations	Merged on Tailgate B618
Inventory	Useless parts in the PDB Left kitting area	Cleared; stock level set for every part
Motion	4 pre-blocks on Roof B618	Same output with 3 pre-blocks
Extra processing	Shields cut 35 mm / 15 mm too long; extra tape issued	Trimmed to drawing; 43 → 39 tape rolls per shift