Value Packs / Manufacturing / Process Control Parameter Optimisation / Wisdom
Which production lines are running more defects than we expect today, and how many units is that across?
NOne line is three times its own ceiling
IN A NUTSHELL
Three lines are running more defects than we expect today. Line 1 is far the worst — 18.4 defects per thousand units against a ceiling of 6.1, which is 758 defective units off that line today. Line 4 is at 8.1 against 6.0 for 232 units, and Line 2 at 7.3 against 6.4 for 269. Lines 3 and 5 are inside what we expect. A usual day on Line 1 gives about 206 defective units.
Defects per thousand units today against what we expect, and the units it cost
Each line is read against its own ceiling — Line 2 is expected to stay under 6.4 and Line 4 under 6.0, because the two run different products at different speeds.
Line 4 is 2.1 defects per thousand over its ceiling and Line 2 0.9 over its own — 232 and 269 units between them. Line 1 is 12.3 over, and 758 units, which is more than the other four lines put together.
Ask a follow-up
How has the defect rate on Line 1 moved day by day against what we expect?
NIt started yesterday and today is worse again
IN A NUTSHELL
Line 1 sat between 3.8 and 6.4 defects per thousand every day from 1 May to 19 May, inside what we expect of it. Yesterday it rose to 12.6 and today it is 18.4 — two days above the range, and today is half again as bad as yesterday. It moved 6.2 overnight against a largest ordinary overnight move of 2.3. The range is nearly level across the three weeks: the line runs continuously, so there is no quiet day to allow for.
Defects per thousand units by day, against the range expected · Line 1
The range is rebuilt each day from the days before it — flat across the three weeks, because a line running 24×7 has no quiet day for the range to bend around.
The two days together are 1,280 defective units where the same two days usually give about 420. Yesterday ran at 12.6 defects per thousand and today at 18.4, 5.8 higher.
Wisdom noticed
Ask a follow-up
Which settings on that line were outside their normal range at the same time?
NTwo settings out of band, at the same time
IN A NUTSHELL
Two settings on Line 1 were out of band over the same two days. Curing zone 3 averaged 181 °C against a normal 166–174 °C, outside it for 31 of the last 48 hours. Wash tank 3 averaged 52 °C against a normal 58–64 °C, outside for 26. Every other setting on the line held inside its range throughout. This is what moved at the same time as the defects. It is not what caused them.
Line 1 settings against their normal ranges, over the last two days
| Setting out of band on Line 1 | Its normal range | Average over the last two days | Hours outside the range |
|---|---|---|---|
| Curing zone 3 temperature | 166–174 °C | 181 °C | 31 of the last 48 |
| Wash tank 3 temperature | 58–64 °C | 52 °C | 26 of the last 48 |
Only the settings that went out of band are listed — the hours column counts how long each sat outside the range the line is meant to hold, out of the forty-eight since the defect rate rose.
Curing zone 3 ran 7 °C above the top of its range and wash tank 3 6 °C below the bottom of its. Both were outside for more than half of the two days. Nothing here shows that either one produced a defect.
Ask a follow-up
Put the line-quality report together.
NThe report lists every batch run on the line over the last two days, with the defects found in it, the material lot it used and the setting readings logged while it ran. One row per batch, so a batch with a bad reading can be found against the units it produced.
line_quality_21_may.xlsx
One row per batch, with its defect count and the settings logged during it. Five lines, and the nineteen days before this started for comparison.
74 KB · 412 rows · 5 lines · 21 days
Ask a follow-up