We shipped a Line 5 to a customer in Rayong last spring, and before the container was even unpacked the owner asked me the same question every buyer asks: when do we hit the catalogue number? The catalogue says 1,200 square meters of finished duct per eight-hour shift. That number is real, but it was measured on a line that had been running the same four sizes for months, with a crew that had done every changeover before, and coils staged in the right order. Nobody buys a line and starts at 100 percent. The distance between day one and rated output is the ramp, and how you manage those first ninety days decides whether you ever actually get there.
I have been through this maybe a dozen times now, on straight lines, U-shapes, and little mini lines in cramped shops. The pattern is always similar. The machine itself is rarely the problem. The ramp stalls on tooling knowledge, material handling, and changeover discipline. So this article is not about the machine spec. It is about what happens after the commissioning engineer flies home.
The temptation in week one is to push the line to full speed to "see what it can do." Resist it. A new crew that has only watched two days of training will feed a coil crooked, the shear will cut out of square, and every jam after that costs more time than the slow running ever did. On the Rayong line we capped the line speed at about 60 percent for the entire first week and told the crew to focus on clean exits, not counts.
Clean exits matter more than most buyers expect. When a blank comes off the line square, with the flange folded flat and the beads crisp, the next operation goes fast. When it comes off twisted, the sheet metal worker spends ninety seconds on the bench fixing it, and your real throughput was lower than if the line had run slower and produced good parts. In week one we tracked first-pass yield, not pieces per shift. If the crew finished a shift with first-pass yield above 92 percent, that was a win, whatever the count.
Before any production run, do a first article on every size you intend to run in week one. Take the blank off the line, check flange height with a gauge, check the corner notch, check bead depth, and measure the diagonal for square. Write the numbers down. That sheet becomes the reference the whole crew works to, and it stops the slow drift where one operator sets the beading slightly shallow and the next one sets it slightly deep.
Then drill the two procedures that cause most early downtime: coil change and size change. Do them deliberately, timed, ten times each if you have to. A coil change on a well-drilled crew takes four minutes including strapping removal and threading. An undrilled crew will spend twenty minutes and bend the leading edge doing it, which then jams at the feed rolls. The size change is worse. On a line with a common rack and motorized adjustments, a drilled crew does a full size change in twelve to fifteen minutes. In week one, expect forty. The difference is not talent. It is whether the person touching the handwheels knows the reference marks and whether somebody wrote the settings down last time.
By the second week the crew knows the line, so this is when we introduce the setup sheet. Every size gets a laminated card: feed length, shear stroke count, beading depth, flange height, servo speed, and the corner insert size. The operator fills in the actual values he used, because the sheet from the factory is a starting point, not the truth. Local material is never identical to the trial coil we ran in the workshop. Zinc weight, yield strength and camber all shift the settings a little.
This is also when we start mixing sizes per shift. A shop that runs one size per day learns the changeover slowly because it happens once. A shop that runs three sizes per shift learns it fast. If your order book allows it, deliberately schedule mixed production in weeks three and four. The target by end of week four is a size change under twenty minutes with no scrap beyond the first blank.
When I time these ramps properly, the distribution surprises people. On a typical early shift, maybe a third of the downtime is changeovers, a third is small jams and misfeeds, and the last third is waiting: waiting for a forklift to bring the next coil, waiting for an inspector, waiting for someone to clear the finished duct off the outfeed table because nobody assigned that job. The third category costs nothing to fix. Assign a deboning and stacking person to the outfeed, stage the next coil before the current one runs out, and you recover twenty to thirty minutes per shift without touching the machine.
After a few of these ramps you can almost predict the problems in order. Here is the usual sequence.
The beading station is the first place a new crew loses quality. Beads come out shallow because the operator is afraid to crack thin galvanized, or the sheet tracks sideways because the side gauges were never squared to the feed. My rule: set bead depth from the material thickness, roughly 3 to 4 millimeters of beading height on 0.8 millimeter sheet, verify it on the first article, and then leave the gauge alone unless the material changes. Flange height drift is the second complaint. TDF flange at 35 millimeters either comes out at 34 or 36 because someone adjusted the forming roll on a hunch. Gauge it, log it, stop adjusting by feel.
The second cluster of problems is upstream of the machine. Coils arrive with camber, or the packing band gouges the edge, or the decoiler was loaded with the coil wound the wrong way and the sheet enters the line with a twist. Cheap coils from a new local mill are the classic ramp killer. The line can process them, but the crew has not learned the compensation yet. I tell buyers to run their first month on the same brand of coil they used during factory acceptance if they can get it, and only then experiment with local material one batch at a time.
If you run two shifts, the second shift is always behind the first, because your best operators gravitate to days. Decide deliberately: either rotate the two strongest people across both shifts for the first month, or accept that night shift ramp lags by two weeks. What kills ramps is letting each shift develop its own settings in secret. One setup sheet, one logbook, both shifts.
Full OEE software is overkill at this stage. A whiteboard and three numbers per shift are enough: planned pieces, good pieces, and total run time lost to stops longer than five minutes. Compute availability as run time over shift time and performance as good pieces over planned pieces, and you get a number the crew can argue about constructively. From the ramps I have watched, a realistic trajectory looks like this:
| Phase | Weeks | Target per shift | Typical bottleneck | First-pass yield |
|---|---|---|---|---|
| Learn the machine | 1 to 2 | 40 to 50% of rated | Jams, misfeeds, operator hesitation | 88 to 92% |
| Lock the setups | 3 to 4 | 60 to 70% of rated | Changeover time, material handling | 93 to 95% |
| Mixed production | 5 to 8 | 75 to 85% of rated | Size changes, coil logistics | 95 to 97% |
| Steady state | 9 to 13 | 90 to 100% of rated | Outfeed clearance, inspection pace | 97%+ |
Those percentages are for a crew that has never touched a line before. If your team came from an older semi-automatic shop, weeks one and two go faster but the changeover discipline still needs the same work, because old habits do not transfer. I have seen veteran sheet metal workers fight the servo feed for two weeks because they trusted their hands more than the encoder.
There is a difference between a ramp problem and a machine problem, and knowing the difference saves everyone money. A machine problem is repeatable and mechanical: the flange fold cracks every sheet at the same corner, or the feed length drifts even after a first article, or a bearing runs hot. Call immediately, with a video of it happening and the material spec you are running. A ramp problem is random and process-shaped: jams that move around the line, yield that dips when a new coil batch arrives, changeover times that vary by operator. Those you fix with the setup sheet and drills, and a good supplier will help you structure that on a video call without charging you a service visit.
Most lines I support send one short video per week during the ramp: a full size change, start to finish. Watching those videos, I can usually tell in the first thirty seconds whether the crew is on track. If the operator is reaching for a reference card, the ramp is healthy. If he is adjusting the beading gauge by eye, I know what the next complaint will be before it arrives.
Can we skip the slow first week if we hire experienced operators?
Experienced at which line? Someone who ran a different brand for years still needs a week on your control layout, your size cards and your material. You can shorten the ramp to maybe six weeks, but you cannot compress week one to two days. The failure mode we see is a confident operator running fast before a first article, then producing two hundred meters of out-of-square duct.
How many pieces of scrap per coil is normal during ramp-up?
In weeks one and two, allow 3 to 5 percent of the sheet for setup blanks and jams. By week eight it should be under 1.5 percent. If scrap stays above 5 percent after a month, the problem is usually material handling or a wrong gauge setting, not operator learning.
Should we run one shift or two during the ramp?
One shift for the first two weeks. Your trainers, the supplier's commissioning engineer and your setup sheet all live on day shift. Once the setup cards are proven and the crew can change sizes without help, split into two shifts. Starting two shifts in week one just doubles the number of undocumented settings you will have to un-teach.
The line met rated speed in testing at the factory. Why is our ramp slower than the plan?
Factory testing runs one size, one coil, one trained operator, no interruptions. Your ramp has to absorb real orders, mixed sizes, local material and a crew of four or five people. The machine is the same machine. The system around it is what takes ninety days to build.