Case studiesRidgeline Container Works · Sample report

Operational baseline

Four-camera observation of fit-up and welding over ten shifts, joined to traveler, quality and scheduling data. Prepared for the president and the VP of operations.

50 days: Median lead time
43 hours: Worked on per canister
3.3 / week: Canisters finished

1. What we found

Ridgeline's canisters spend 50 days on the floor for 43 hours of work. The constraint is fit-up, not welding. The welders wait on the fitters for about a tenth of the shift, and the manual booths have the arc lit 18 to 26 percent of the time. Five changes that need no capital raise throughput from 3.3 to about 4.5 canisters a week.

Flow. A canister is being worked on 3.6 percent of the time it is in the building. Of the 50-day lead time, 37 days are queue, and the two largest queues are both in front of fit-up.
Constraint. Fit-up is loaded at 105 percent of its available hours. Fit-up Station A is hands-on for 68 percent of the shift. The rest is crane waits, inspector waits and walking, so the constraint is under-used before it needs more capacity.
Welding. Shift start, shift end and the minutes after each break account for much of the gap between the booths and a well-run manual station.
Method. The same circ seam takes a median of 45 minutes by the fastest observed method and 68 by the slowest. The quality of the fit-up received moves the median from 42 to 60 minutes.
Layout. A canister takes 13 crane lifts and about 1,100 feet of travel. Welders walk to the consumables crib about 24 times a shift.
Automation. On today's numbers, automating welding without relieving fit-up adds no throughput.

2. Data and method

The analysis joins three kinds of data: what the cameras saw, what the travelers and the ERP recorded, and what the quality system logged. Each minute of video at each station was classified into an activity, and the activities roll up into five categories: value-added work, support work, waiting, walking, and breaks and idle.

Activity classification: Coverage4 cameras, per minute, 10 shifts; SourceCobalt video pipeline
Traveler events: Coverage24 canisters × 10 operations; SourceERP job costing and sign-offs
Per-joint weld cycles: Coverage141 observed joints; SourceCamera events and traveler joint IDs
Load against capacity: Coverage9 work centers, next 8 weeks; SourceERP scheduling and headcount
Daily work in process: Coverage9 locations × 10 days; SourceCamera count and floor walk
Nonconformance log: Coverage12 months; SourceQuality system

Dataset	Coverage	Source
Activity classification	4 cameras, per minute, 10 shifts	Cobalt video pipeline
Traveler events	24 canisters × 10 operations	ERP job costing and sign-offs
Per-joint weld cycles	141 observed joints	Camera events and traveler joint IDs
Load against capacity	9 work centers, next 8 weeks	ERP scheduling and headcount
Daily work in process	9 locations × 10 days	Camera count and floor walk
Nonconformance log	12 months	Quality system

Before the classification was used, it was checked class by class against an industrial engineer who coded a sample of the footage independently. Grinding as preparation and grinding as rework are the hardest to tell apart, so rework figures are quoted as a range and were spot-checked against traveler joint IDs.

3. Current state: how a canister flows

The value stream map is the one-page picture of the plant. The boxes are the operations on a canister's traveler, the triangles are canisters waiting between them, and the two rungs underneath split the time into waiting and being worked on. The two highlighted boxes are the same work center, fit-up, and the largest triangles sit in front of them. The arithmetic checks: 23 open canisters at 3.3 a week is about 49 days, which is the measured lead time.

Current-state value stream map

One canister family. Median queue before each step, and hours worked at it.

2.0 d
Cut and form3 h
1.5 d
Roll2 h
2.5 d
Long seam4 h
7.0 d
Fit-up: heads6 h
3.0 d
Circ seams9 h
8.5 d
Fit-up: nozzles5 h
2.5 d
Nozzle welds7 h
5.0 d
Radiography2 h
2.5 d
Pressure test2 h
2.5 d
Finish3 h

WaitingWorked onConstraint: fit-up

Lead time: 50 days
Queued: 37 days
Worked on: 43 hours
Share of lead time worked on: 3.6%

4. The constraint

Three independent signals point to fit-up. The floor count shows where canisters pile up. The ERP load shows which work center has more scheduled hours than it has people. The cameras at the weld booths show welders standing in an empty booth waiting for a fitted assembly.

Load by station

Scheduled hours as a share of available hours. Under each bar, the units queued in front of that station.

62%

58%

105%

81%

94%

66%

2 queued

Cut and form

1 queued

Roll

7 queued

Fit-up

3 queued

Weld

4 queued

NDE

2 queued

Finish

Bottleneck: Fit-up. It is loaded above its available hours and has the deepest queue.

Roughly a fifth of the constraint's day is waiting or walking. An hour recovered at fit-up is an hour of plant throughput. The same hour recovered at the long seam station is worth nothing today. This ordering drives the ranking in section 7.

Idle by cause

Fit-up Station A, minutes per shift

Waiting for the crane

Walking to the tool crib

Waiting for an inspector or a drawing answer

Waiting for consumables

5. What each station is doing

The time-use chart is the camera equivalent of a classic work sampling study, with about 4,600 observations per station in place of a few hundred.

Time use by station

Share of shift, ten shifts pooled

Value-added work
Support work
Waiting
Walking
Breaks and idle

Fit-up Station A

39%

29%

13%

Weld Booth 3

18%

38%

17%

11%

16%

Weld Booth 5

26%

37%

14%

SAW long seam

31%

36%

12%

15%

Time use by station, percent of shift
Station	Value-added work	Support work	Waiting	Walking	Breaks and idle
Fit-up Station A	39%	29%	13%	6%	13%
Weld Booth 3	18%	38%	17%	11%	16%
Weld Booth 5	26%	37%	14%	9%	14%
SAW long seam	31%	36%	12%	6%	15%

Shift profile

Share of covered stations with hands-on work under way, by half hour

Hands-on work by half hour, percent of covered stations
Time	Hands-on
06:30	31%
07:00	58%
07:30	71%
08:00	74%
08:30	72%
09:00	38%
09:30	61%
10:00	73%
10:30	75%
11:00	69%
11:30	22%
12:00	49%
12:30	68%
13:00	70%
13:30	41%
14:00	57%
14:30	29%

The losses that are easiest to recover, at shift start, after breaks, walking to the crib and waiting for consumables, are not welding problems. They are staging problems, and one material handler solves most of them.

6. Method variation and standard work

Differences in method between stations are stable from day to day, which means they can be standardized. Standard work is not a quality procedure: the weld specification already governs amps, volts and travel speed. It is the sequence and staging of the fastest, cleanest observed method, written down so that every station can use it.

Standard work: fit up and tack, head to shell

Best-observed method, Fit-up Station A. Reviewed with the station's leads.

Steps: 6
Standard time: 41 min
Revision: A

Step 13 min
Confirm the drawing revision and weld map
Stop if the revision does not match the traveler.
Step 26 min
Stage the shell, head and tack consumables
Staged by the material handler before the shift, not by the fitter.
Step 38 min
Set the shell on the rollers and level it
Level before the head is lifted. Re-levelling under the crane holds the crane.
Step 412 min
Align the head and check the gap
A gap over 1/8 in goes back to forming. It is not welded up.
Step 59 min
Tack at four points in opposite pairs
Re-check the gap after the second tack.
Step 63 min
Sign the traveler and call the hold point
Call the inspector now, so the booth is not left waiting.

7. Changes, ranked by throughput

The ranking follows the constraint. Anything that recovers fit-up hours, or keeps a poor fit-up out of the booths, comes first.

Ranked changes

Estimated value in canisters per week

1
Stage kits and consumables with one material handler
Fit-up, weld booths
Cam 1 · 07:12Cam 3 · 09:41
+0.4 / week
2
Book the hold-point inspector when fit-up starts, not when welding ends
Fit-up
Cam 1 · 10:26
+0.3 / week
3
Check the fit-up gap before the assembly leaves the station
Fit-up
Cam 1 · 13:05Cam 4 · 13:48
+0.2 / week
4
Adopt the best-observed circ seam sequence at all four booths
Weld booths
Cam 3 · 08:15Cam 4 · 08:22
+0.2 / week
5
Stage the first job of the shift the afternoon before
All stations
Cam 2 · 06:34
+0.1 / week

Throughput model

With changes 1 to 5

Cut and form

78%

Constraint

Roll

73%

Constraint

Fit-up

92%

Constraint

Weld

96%

Constraint

NDE

100%

Constraint

Finish

83%

Constraint

Load: scheduled hours as a share of available hours.

Throughput: 4.5canisters per week+36% against baseline
Constraint: NDElimits the whole lineMoves from Fit-up
Cost per unit: $17,900labor and overhead-16% against baseline

Two cautions. The effects are only approximately additive: once fit-up is relieved the constraint moves, most likely to the radiography windows, and the ranking should be redone at that point. And the first change assumes a helper can be reassigned. If not, that role is the first hire to make, ahead of any welding automation.

8. Baseline for judging welding automation

A robotic cell judged against today's average will look better than it is. It should be judged against the best-observed manual cycle, at the same inspection method, with rework and setup counted on both sides, and only once fit-up is confirmed as no longer the constraint.

Arc time as a share of the shift: Current baseline18% and 26%, manual booths; How it is measuredContinuous, per station
Cycle time per circ seam: Current baseline45 to 68 min, by method; How it is measuredPer joint, timestamped
First-pass yield at radiography: Current baseline84%; How it is measuredWeekly, from the NDE vendor
Rework per canister, observed: Current baseline4.5 hours; How it is measuredCamera: grind-out and re-weld
Setup per canister at the booth: Current baseline38 min, median; How it is measuredCamera: last cap to first arc
Canisters finished per week: Current baseline3.3; How it is measuredERP completions

Measure	Current baseline	How it is measured
Arc time as a share of the shift	18% and 26%, manual booths	Continuous, per station
Cycle time per circ seam	45 to 68 min, by method	Per joint, timestamped
First-pass yield at radiography	84%	Weekly, from the NDE vendor
Rework per canister, observed	4.5 hours	Camera: grind-out and re-weld
Setup per canister at the booth	38 min, median	Camera: last cap to first arc
Canisters finished per week	3.3	ERP completions

The same cameras stay in place. A robotic cell would get a camera of its own, with the same activity classes and two more for programming and fixture loading, and the comparison would run over twenty shifts on the same product mix.

Assumptions and limits

One shift was observed. Second-shift behaviour is unmeasured.
Cameras covered one of two fit-up stations and two of four booths. Station-level figures are extrapolated to the unobserved stations for the throughput model.
Hours worked come from job costing, which operators book at the end of the day and round. Elapsed camera time was used wherever the two disagreed.

See it on your own floor

This is the report for a fictional plant. Get in touch to talk about one for yours.