Fab shops have the most chaotic schedules in manufacturing
A fabrication job might touch six different machines. Raw material gets cut on the saw. Flat stock goes to the brake for bending. Parts get tacked, then welded. Welds get ground smooth. Everything goes to the paint booth. Then inspection and shipping.
Each step uses a different machine. Each machine has a queue. Each queue has conflicts with every other job moving through the shop. One bottleneck at the welding station backs up everything behind it.
That’s the scheduling problem in metal fabrication. It’s not “put a job on a machine.” It’s “route 15 jobs through 8 machines in the right order without anything crashing into anything else.”
What makes fab shop scheduling different
Diverse machine types. A CNC shop might have 5 mills and 3 lathes. A fab shop has a plasma cutter, a shear, two brakes, a welding bay, a grinder, a paint booth, and a crane. Each machine is different. The schedule has to handle all of them.
Long operation chains. A typical fab job has 4 to 8 steps. Cut, form, weld, grind, finish, inspect. Each step depends on the previous one finishing. If the saw falls behind, every downstream operation shifts.
Shared bottleneck machines. The welding station is usually the bottleneck. Every job passes through it. If three jobs need welding on Tuesday, two of them are waiting. Your schedule needs to show that conflict before Tuesday arrives.
Variable job sizes. One job is a single bracket that takes 20 minutes. The next is a structural frame that takes three days. Both go through the same machines. Scheduling has to handle both without one blocking the other.
Material variety. Steel, aluminum, stainless, galvanized. Different gauges, different lengths. You need to know what’s on the rack before you schedule a job.
Scheduling the line, not the machines
The mistake that costs fab shops the most is scheduling each machine well and the line badly.
You can fill the plasma table for a solid week and still ship late, because a fab job is a chain and a chain moves at the speed of its slowest link. The parts you cut on Monday do nothing for you until they have been bent, and the bend does nothing until the welder is free. Every hour of finished cutting that has nowhere to go is just inventory sitting on a cart.
Line scheduling means sequencing the chain rather than each station:
Work backwards from the constraint, not the due date. Find the machine everything queues at. In most fab shops it is the welding bay, sometimes the brake if you run a lot of formed sheet. That machine’s queue is your actual production plan. Everything upstream exists to keep it fed and everything downstream exists to clear what it finishes. If you have never confirmed which machine that is, find the real bottleneck first, because shops are wrong about this more often than not.
Do not start jobs as early as possible. It feels productive and it is the single biggest source of floor clutter. Releasing a job to the saw three weeks before the welder can take it produces a pile of cut steel, a tote that has to be stored and found again, and a job whose material is committed while a hotter order needs the same plate. Release work so it lands at the constraint shortly before the constraint needs it.
Schedule the handoffs, not just the operations. The gap between operations is usually longer than the operations. A part waits on a cart for the welder, waits again for the grinder, waits again to go in the booth. Those queue hours are most of your lead time even though none of them appear on a machine. If your schedule only shows the run hours, it is describing a job that finishes days before the real one does.
Assume the chain will break and plan the recovery, not the perfect run. A weld fails inspection, plate arrives short, a customer pulls a date in. In a chain, one change re-sequences everything downstream of it. The value of a live schedule is not that it predicts the week correctly. It is that when Tuesday goes wrong you can see in a minute which four jobs just moved and which customer to call.
Getting the schedule onto the shop floor
A fab shop schedule fails in one specific place: between the office and the bay.
The plan gets built on a computer, printed on Monday, and taped to a post. By Tuesday afternoon a rush job has landed, two operations have been resequenced, and the sheet on the post is describing a week that no longer exists. The welder is still working from it, because it is the only thing in front of them.
So the office knows one schedule and the floor runs a different one. Everything after that is downstream of that split: the status meeting where nobody agrees on where a job is, the customer call answered with a guess, the part that gets ground twice because nobody marked the first pass.
Shop floor scheduling means the plan and the floor are the same object:
The person at the machine sees what is next. Not the whole shop, not a Gantt chart they have to interpret. The welder needs their own queue in order, with the job in front of them at the top.
Status changes where the work happens. A job marked complete at the station is worth more than the same update entered by the office an hour later from a scribbled note, because in between those two moments everyone else was looking at a schedule that was wrong.
A phone counts as a terminal. Fab shops do not want a PC at the welding bay, and they should not need one. Your welders already carry a screen.
Paper can stay if it wants to. Plenty of fab shops will keep running a traveler with the job and that is fine, including what it should carry. What has to stop is the traveler being the only place a job’s status exists, because then the office cannot answer a question without walking the floor. Keep the paper, mirror the status. See tracking jobs across multiple machines for how that works in practice.
Batching, nesting, and the fight with due dates
Every fab shop batches, and batching is where a schedule quietly goes wrong.
You nest six jobs onto one sheet of 10 gauge because the drop is expensive and the plasma setup is the same either way. You group every 90 degree bend in the same tooling so the brake gets set once instead of four times. On its own each of those decisions is correct. Setup is real money and changeover eats more capacity than most shops realise.
The problem is what batching does to dates. A nest is only as urgent as its most relaxed job, and the moment you combine a Friday job with three jobs due the following month, you have either pulled three jobs forward for no reason or you are holding the Friday job until the batch is worth running. One of those wastes material and floor space. The other misses the date.
Batch scheduling means deciding that trade deliberately instead of discovering it:
Batch by setup, not by convenience. Grouping is worth doing when it removes a genuine changeover. Grouping parts that share a material but need different tooling saves you very little and costs you sequencing freedom.
Put a ceiling on how far a batch can pull work forward. A useful rule is that a job only joins a nest if it was going to be cut within the next week or two anyway. Beyond that the material cost you saved comes back as carrying cost, congestion, and cash tied up in plate.
Never batch through the constraint. Upstream of the welding bay, batching is mostly free because you have slack. At the constraint, an hour is an hour and a large batch means the jobs behind it wait for all of it. Batch the saw and the plasma. Sequence the bay one job at a time.
Watch the pattern where the same customer is always late. Very often it is the customer whose parts are small and awkward, so they never justify a nest on their own and keep getting held for a batch that never quite forms.
Sizing the bottleneck honestly
Capacity planning in a fab shop comes down to one number: how many hours the constraint really has in a week, and how many you have already promised.
Most shops get this wrong in the same direction. The welding bay has two welders and the shop runs 40 hours, so the plan assumes 80 hours of welding a week. What the bay actually delivers is closer to 55, once you subtract fit-up, tacking, repositioning, waiting for the crane, the inspection that comes back, and the two hours on Friday that go to whatever caught fire. Load the bay to 80 and it will fail every week without anyone being able to say why.
Two things fix it:
Time the whole operation, door to door. Not the arc time. The hours between the part arriving at the bay and leaving it, including everything that is not welding. If you schedule from arc time you are booking capacity you do not have, which is why jobs run late even when nothing broke.
Refuse to book past the real number. A schedule that lets you put 80 hours into a 55 hour bay is not a schedule, it is a wish list with a grid on it. This is what finite capacity scheduling means, and it is the difference between finding out about an overload on Monday while you can still subcontract, and finding out on Thursday when you cannot. Most tools plan the opposite way without telling you, which is the whole gap between finite and infinite capacity.
Get those two right and quoting fixes itself, because the date the board gives you is a date the shop can hit. That is how you quote lead times you can defend instead of padding every job by two weeks and losing work to a shop that quoted straight.
How Machestra works for fab shops
Every machine on one board. Saw, brake, welder, grinder, paint booth. All on the scheduling board. You see what’s running where and when each machine opens up.
Per-operation scheduling. Break the job into steps: cut on the plasma, bend on the brake, weld at station 2, grind, paint. Each step gets its own machine, operator, and time slot. You always know where a job is in the process.
Conflict detection. Three jobs need the welding station on the same day? Machestra shows you the overlap before it becomes a problem on the floor. No more double-bookings.
Material tracking. Track your sheet stock, bar stock, and tube by type, gauge, and quantity. Get low-stock alerts before you schedule a job you can’t finish.
Mobile access. Your welder checks the schedule from a phone at the station. Your office staff checks from the desk. You check from home. Same data, everywhere.
One thing it deliberately does not do: sequence the jobs for you. Machestra will not let you book a machine or a person twice and it will show you the overload while you can still act on it, but the order the work runs in stays your call. In a shop this size you already know which customer waits and which one does not, and no solver is going to learn that from a routing file.
What you don’t need
You don’t need CAD integration. You don’t need EN 1090-2 compliance tracking. You don’t need a quoting module or an accounting system bolted onto your scheduling tool.
If you do need those things, Eziil is built for that. It costs 5x more, but it covers the compliance angle.
If you need scheduling and nothing else, Machestra stays out of your way. Four things: machines, people, materials, schedules.
Other shops look at MRP systems like MRPeasy, but at $49/user/month the cost adds up fast when you want your whole fab crew on the schedule.
Ready to try it? Start free with 3 machines →. Set up in under 10 minutes. No credit card.
Pricing
No per-user fees. Your welders, fitters, office staff, and shop foreman all see the schedule.
- Free: 3 machines, 3 users. No credit card.
- Starter ($19/mo):* 10 machines, 50 users.
- Growth ($79/mo):* 100 machines, 100 users.
A 6-person fab shop with 8 machines pays $19/mo* total. Not per user. Total.
Getting started
- Sign up free
- Import your machines and jobs from your spreadsheet
- Start scheduling across your whole shop
Under 10 minutes from signup to your first scheduled job.
*All Machestra prices shown in USD. Actual price may vary based on your location.