Reconfigure into U-cells, lift work off the floor
Throughput

Reconfigure into U-cells, lift work off the floor

Financial impact

$7,706USD/ yr EBITDA

Operational impact

1.5operators freed up
high confidence

The math

Gross labor recovery+$13,440

(1 + 0.5) FTE × 2240 hrs × $4/hr

Multi-skill pay reinvestment−$5,734

8.0% × 8 operators × 2240 hrs × $4/hr

Net annual value$7,706/ yr

gross minus reinvestment

Capex

$1,500

Payback

2 months

Difficulty

3 / 5

Time to implement

~4 weeks

Full-time workers freed by evening out the line
Full-time workers freed by ending the bending to floor bags
Cross-training pay raise
%
Workers getting the cross-training raise
One-time setup cost (bins, mats, tape)
$

Also uses working days / yr 280 days, loaded labor rate $4/hr. Edit on the dashboard.

Description

Today the sewing machines sit in long straight rows, and each operator runs one machine doing one task over and over. Parts move between them in tied bundles, one single-task worker to the next (the traditional "bundle" setup). The rows are badly out of balance: one part of the floor shows six machines with only three staffed, so faster stations pile up half-finished garments in front of slower ones. That half-finished work overflows the tables into brown paper bags sitting on the concrete, so operators bend to the floor for every piece. The result is a lot of walking, waiting, and bending that never turns into a finished garment.

Evidence

  • [00:01:00] Six machines in the front-left bank, only three staffed: the line is badly out of balance
  • [00:00:46] Half-finished garments stuffed into paper bags on the concrete floor
  • [00:01:29] Empty wooden stools and work piling up between unstaffed machines
  • [00:00:30, 00:01:29] Machines on freestanding, unbolted H-frame tables, which is why the fix is cheap
  • [00:00:58] Power cables zip-tied to the center racks, which is why Card 3 has to come first

Deep dive

Expand for detail

Load-bearing, card breaks if wrong

The racks are already out and power comes from the ceiling (Card 3 is done).

Today the machines' power is zip-tied to the steel racks running down the center of the floor, visible at [00:00:58]. Before any machine can pivot into a U, those racks have to come out and the power has to drop from the ceiling instead. That is Card 3, and it is a hard prerequisite: until it is done, the tables are tethered and cannot move. What it looks like on the floor if this is wrong: you start pivoting a table and the power cords pull taut, or the electrician says the ceiling drops are not in yet and the new cell cannot be switched on.

Enough of the weekly order mix repeats to keep a cell balanced.

A U-cell is fastest when the same kind of garment runs through it long enough for the team to find its rhythm and keep the stations balanced. This is a build-to-order shop (it only makes what is already ordered, and the product can change week to week). If next week's order is a different garment with a different sequence of operations, the supervisor is re-balancing the cell every morning, operators never hit stride, and a flexible cell can actually finish LESS than a well-run straight line. This is the single strongest reason the card could underdeliver. What it looks like on the floor if this is wrong: the supervisor spends the first hour of every shift reassigning people, and the cell's output swings wildly day to day. The fix, if the mix is choppy: build the cell only around the repeatable 60 to 70% of volume and leave the oddball orders on the old straight-line stations.

Operators will run 2 to 3 machines for a raise capped near +8.7%.

The whole gain survives only if operators will learn to run two or three machines for a raise of no more than about 8.7% (roughly $0.35 an hour, taking $4.00 up to about $4.35 loaded). Sample and multi-machine operators in Dubai can command $5.50 to $6.00 an hour. If workers hold out for that, the raise eats the entire labor savings and the card goes negative. The lever that makes a small raise stick is paying it as a team bonus on finished pieces, not a flat hourly bump. What it looks like on the floor if this is wrong: operators grumble about 'doing three jobs for one wage' and quietly slow down, or the ones you cross-train quit for a shop that pays the higher machine rate.

The freed-up worker time actually leaves the payroll.

Recovering about 1.5 workers' worth of time only becomes profit if those hours actually leave the cost base: either the freed people move to the real bottleneck (pressing, Card 1) to create more shippable output, or the headcount is allowed to fall through normal attrition without hiring replacements. If the freed workers simply stay on as 'helpers' with nothing to do, the floor is calmer but the wage bill is identical and the savings never reach the bottom line. What it looks like on the floor if this is wrong: 30 days after the change the daily headcount sheet is unchanged and one or two people are visibly under-occupied.

Supportive, card weakens but survives

The tables are freestanding and the floor fits the cells with safe walkways.

Two physical facts keep this cheap and are already confirmed on the walkthrough. The sewing machines sit on freestanding, unbolted H-frame tables (the plain metal leg-frames shaped like an H), seen at [00:00:30] and [00:01:29], so rearranging them costs labor and a few bins, not new machinery. And the sewing zone measures about 17.6 by 8.6 meters, which comfortably fits four to five cells of five machines while keeping a 0.9 meter (36 inch) clear exit aisle, the minimum width a person needs to get out safely, once the center racks are gone. What it looks like on the floor if this is wrong: a table turns out to be bolted down or hard-plumbed, or the taped-out cells leave the walkway under 0.9 meters and Dubai Civil Defence flags the blocked exit path.

The freed-up workers just stay on the payroll

Risk

The cell works and gives back about 1.5 workers' worth of time, but with no labor-cost discipline those people stay on as 'helpers' or floaters. The floor is calmer, the wage bill is unchanged, and the savings never reach profit.

Early Warning

30 days after launch the daily headcount sheet is unchanged, and one or two people are visibly under-occupied.

Mitigation

Decide up front where the freed time goes: move it to the real bottleneck (pressing, Card 1) to create more shippable output, or let headcount fall through normal attrition without hiring replacements. Freeze backfill hiring until the freed time is placed.

Stake

All of the roughly $14K a year.

A choppy order mix forces daily re-balancing

Risk

This is a build-to-order shop, and a cell is only fast when the same garment runs long enough to find a rhythm. If the product changes constantly, the supervisor re-balances the cell every morning, operators never hit stride, and the cell can finish less than the old straight line did. This is the single most likely reason the card underdelivers.

Early Warning

The supervisor spends the first hour of most shifts reassigning people, and the cell's daily output swings sharply.

Mitigation

Build the cell only around the repeatable 60 to 70% of volume (a 'flex cell') and leave the oddball orders on the old straight-line stations. Time-study the top 3 to 4 garments first so the balance holds for the bulk of the work.

Stake

Most of the gain; a badly matched cell can erode it toward zero.

Operators reject the raise or the multi-machine work and slow down

Risk

Workers may see a small raise for running three machines as insulting ('three jobs for one wage') and quietly slow down, or the skilled ones leave for a shop paying full multi-machine rates. Either way the labor recovery evaporates and you may lose trained people.

Early Warning

Cell output sits 20% or more below the old batch rate by day three, or one or two skilled operators quit right after the change.

Mitigation

Pay the raise as a team bonus on finished pieces, not a flat hourly bump: hit the target and the team earns the +8%, beat it and they can earn more. Start with willing operators and keep a reluctant one as a single-task anchor.

Stake

The full gain, plus a transition dip and the cost of replacing anyone who quits (retraining one operator can exceed a month of the savings).

Discipline slips and the floor reverts to batch flow

Risk

Two old habits kill the cell if they creep back. Someone gets a big batch, overflows the bin, and drops a paper bag on the floor again [00:00:46]; within two weeks bags are the default and the bins are junk storage. Or operators drag the heavy fixed wooden stools [00:01:29] back in, batch 30 pieces at one machine, then stand up and move, which destroys the piece-by-piece flow the cell depends on.

Early Warning

A single paper bag appears near the cell, or a stack of more than about five half-finished pieces builds up on one machine table.

Mitigation

Zero tolerance on paper bags; confiscate them factory-wide. Physically remove all rigid seating from the cell so only rolling stools or standing mats remain.

Stake

The gains from evening out the line and ending the bending, about $13K.

A. How to build the first cell

  • Where: the front-left bank [00:01:00]. It holds six machines but only three are staffed, which is the clearest proof of the imbalance, and it has a clear line of sight for the supervisor and easy access to the center aisle.
  • Which machines, and what each does: a five-machine U of three basic straight-stitch machines plus one edge-finisher plus one specialty, matching what the floor already owns (41 single-needle lockstitch machines, the plain straight-stitch workhorse, plus 6 overlockers, the machines that clean and trim a raw seam edge in one pass).
    • Station 1 (Juki straight-stitch): join the first two panels together. Run by operator A.
    • Station 2 (Typical overlock): clean and trim the raw seam edge so it will not fray. An overlock, also called a serger, stitches and cuts the edge in one pass. Operator A.
    • Station 3 (Brother straight-stitch): the visible outside stitch line. This is the shared hand-off: operator A drops the piece in the bin here, operator B picks it up.
    • Station 4 (Juki straight-stitch): pockets and small sub-assembly. Operator B.
    • Station 5 (Juki straight-stitch): close the last seams. Operator B.
  • The bins: 600 by 400 by 320 mm heavy-duty stacking containers (Fami or a local Dubai equivalent) on wheeled steel bases, so the small buffer of half-finished pieces sits at waist height, about 80 cm, instead of on the concrete. About 250 AED (~$68) per set.
  • The floor markings: 50 mm 3M 471 yellow floor tape, marking a strict 3 by 2.5 meter box. No half-finished work, no bags, and nobody not assigned to the cell crosses the tape.

B. How to pay for the cross-training without eating the savings

Pay for skill, and pay it as a team bonus on finished pieces, not as a flat base-pay raise. The whole card depends on keeping the fully loaded rate at or below about $4.35 an hour. Think of three simple skill steps:

Skill stepMachines the worker can runRaiseExtra pay a year, eachWorkersCost a year
Trainee1 machine0%$0most of the floor$0
Versatile2 machines+4.0%$3584$1,432
Cell lead3 machines+8.0%$7166$4,296
Total~$5,728 / yr

The net, honestly: the cross-training raises across the floor cost about $5,728 a year. Set against the roughly $13,440 of labor the balanced cells free up, that nets to about $7,712. The roughly $14K on the dashboard is the gross labor recovered; the live model is deliberately more conservative because it subtracts the raises, which is exactly why the raise must stay capped near 8.7%. Push the raise to full multi-machine wages ($5.50 to $6.00 an hour) and the reinvestment overtakes the recovery and the card goes negative.

C. Why this cheap rearrangement, and the one thing that can beat it

  • Why not just leave the straight rows alone? The straight-row setup (the "bundle" system, where parts move in tied bundles from one single-task operator to the next) maximizes how many times each person repeats one operation, but it piles up huge stacks of half-finished work between stations and sends every piece on a long round-trip. That is exactly the scatter of freestanding tables and floor bags the walkthrough shows. Cells cut the travel and the piles.
  • Why not just cross-train people without moving the tables? Because the imbalance and the bending are baked into the layout. The U shape is what lets one person tend two or three machines standing in one spot, and the waist-high bins are what end the bending. Cross-training on a straight row recovers a fraction of the gain.
  • Why not buy new machines or automation? Nothing here is a machine problem. The machines are fine; the waste is in the LAYOUT and in one-person-one-machine staffing. About $1,500 of bins and tape beats any capital purchase, and the tables are already freestanding, so moving them costs labor, not hardware.
  • The one thing that can beat it (the honest catch): this is a build-to-order shop with a product that can change week to week. A cell shines when the same garment runs through it long enough to build rhythm; a constantly changing mix forces daily re-balancing and can make the cell slower than a well-run straight line. The protocol: time-study the top 3 to 4 garments, build the cell only for the repeatable 60 to 70% of volume, keep straight-line stations for oddball orders, and scale to a second cell only after the pilot beats the old baseline for 3 to 4 weeks straight.

What we did not have in the corpus and would request from the client on day one. The aggregated gaps across all 15 cards become the engagement-letter ask sheet.

  1. 01

    Payroll register (base pay plus allowances, per operator)

    Owner

    HR / Finance

    The exact wage make-up per operator. This is the number the whole card pivots on: if base pay is under $4 the return grows, but if the cross-training raise pushes the loaded rate past about $4.35 an hour the card turns negative.

  2. 02

    8 to 12 weeks of order history

    Owner

    Sales / Production

    Measures how much the product mix changes week to week. This is the load-bearing test of whether a cell can stay balanced. If the mix is too choppy, scope the cell down to the repeatable 60 to 70% of volume.

  3. 03

    Per-station time study (last 30 days, by operation)

    Owner

    Production Manager

    The real time each operation takes, so the five pilot machines can be balanced and operator A does not starve or flood operator B.

  4. 04

    Output per operator-hour by garment type (the current baseline)

    Owner

    Production Manager

    The straight-line number the pilot cell has to match or beat for 3 to 4 weeks before any money is spent scaling.

  5. 05

    Turnover rate (trailing 12 months)

    Owner

    HR

    If turnover already runs above about 40%, the floor may be too unstable to cross-train: you would train people who leave for higher-paying machine jobs before the change pays back.