How Many Machines Does The Job Actually Need?
2026/11/17
How Many Machines Does the Job Actually Need?
The question that arrives by email is almost always "which machine should I buy". The question underneath it is "how many", and it is answered first, because one machine working at 95 percent utilisation and two working at 60 percent are different purchases, different floor layouts and different amounts of grief.
The arithmetic takes ten minutes and needs four numbers. Most of the errors come from the two factors that get left out rather than from the sums.
A hall of press welders at the end of assembly. Whether a job needs one of these or four is arithmetic, not opinion, and the arithmetic starts with the part.
Four Numbers Before Any Machine
Parts per shift, from the production plan rather than the best week. Welds per part, counted on the drawing and then checked on a real part, because brackets and clips appear on drawings that nobody welds. Seconds per weld, measured with a stopwatch on a comparable machine including the index between welds. And handling time per part: pick up, locate, clamp, release, stack.
The last one is the number buyers underestimate most often, because it does not appear in any machine specification. On a four-weld part it can easily be a third of the total.
The Arithmetic, and the Two Factors People Leave Out
Multiply welds by seconds per weld, add handling, and you have the cycle time per part. Multiply by parts per shift and you have the machine-seconds the shift demands. Divide by the seconds actually available in a shift and you have the number of machines.
The first factor people leave out is availability. A 7.5-hour shift is 27,000 seconds, but no line runs at 100 percent: electrode dressing, changeovers, a part that jams, a break. Planning at 80 percent is standard and 75 percent is honest. The second factor is the machine's duty cycle at the working point, which is a different calculation and is covered below.
A Worked Example
Take a bracket with four welds, six seconds each including the index, and eight seconds of handling. The cycle is 32 seconds per part. At 900 parts per shift the demand is 28,800 seconds, against 21,600 usable seconds in a shift at 80 percent availability. That is 1.33 machines, so the answer is two, each running at about 67 percent.
Parts per shift |
Machine-seconds needed |
Machines at 80% availability |
Comment |
600 |
19,200 |
One, at 89% |
Workable, but no room for a bad day |
900 |
28,800 |
Two, at 67% |
Comfortable, and one can be serviced |
1,200 |
38,400 |
Two at 89%, or three at 59% |
Two is tight; the third machine buys uptime |
The table is the whole argument. The buyer who plans at 100 percent buys one machine for the 600-part case and discovers in month two that a single tip change costs the shift its target.
Where the Seconds Hide
Three things sit outside the cycle time and inside the shift. The first is electrode dressing: a manual dressing every few hundred welds is minutes, and those minutes are production. The second is changeover, which on a family of parts can be the largest single consumer of capacity and is invisible in a per-weld calculation. The third is the first hour of a shift, when the machine is cold, the operator is finding the fixtures and the first parts are being checked.
The last thing outside the cycle is what the shift does at the end of it: the count, the wipe-down, and the tool change that gets deferred to the morning. An hour a shift is more than ten percent of capacity, and on a machine planned at 89 percent that is the whole difference between finishing the plan and explaining why it did not finish.
An automated line with two welding stations on a common conveyor. Automation does not change the arithmetic; it changes which seconds are inside the cycle and which are not.
The Duty Cycle Trap in Capacity Planning
A spot welder rated at 50 percent duty cycle can weld for five seconds in every ten and no more, at its rated current. On the four-weld bracket above, welding occupies 24 of every 32 seconds — 75 percent — so the machine is not being asked to run at 67 percent of its capacity, it is being asked to run at 75 percent of its duty cycle before any allowance for the next part arrives.
That mismatch shows up as a machine that trips on a warm afternoon, or as a schedule that has to be slowed after the first hour. The fix is a larger machine, a lower-current schedule with a longer weld time, or a different process. All three are cheaper to decide before the purchase order.
What Changes When the Part Changes
Every capacity calculation is done on one part and then used for a family. The number that decides whether that is safe is changeover: how long it takes to go from the four-weld bracket to the six-weld bracket in the same family. If the fixture needs a new nest and the controller a new schedule, the changeover is twenty minutes and the calculation above still holds. If it needs the toolroom, the changeover is a day and the calculation is fiction.
The useful test is to take the part the customer orders least often and run the arithmetic for a batch of thirty. On a machine planned at 67 percent that batch fits. On a machine planned at 89 percent it does not, and the line stops for a day to make a week of sales.
Automation Moves the Seconds; It Does Not Remove Them
A robot cell removes the handling time, which on a four-weld part is a third of the cycle. What it adds is a longer time per weld, because the arm has to travel between them, and a loading regime, because parts arrive in a nest and somebody still has to fill it. The gain is real and it is rarely the factor the vendor implies.
It also changes the failure mode of the calculation. One cell down is not a machine's worth of capacity lost at the margin; it is the whole output of that cell, and the buffer in front of it is whatever the conveyor holds. That is why cells are often bought in pairs, and why a single cell is frequently paired with a manual press as a fallback.
The Capacity You Cannot See
Two machines at 67 percent is not the same as one machine at 134 percent, and the difference is not arithmetic. It is that either machine can be stopped: for a tip change, for a service, for the afternoon when the chiller is struggling. That is the whole reason to buy the second one, and it never appears on a sheet that divides demand by capacity.
The same logic covers growth. If the plan is to add a second customer for the same part family in eighteen months, the 600-part case that needs one machine at 89 percent is the worst of both answers: no slack today and no room tomorrow.
One Machine Too Few, One Too Many
One machine too few is the expensive mistake. It shows up as overtime, as a second shift added for capacity rather than demand, and as a maintenance backlog, because a machine at 95 percent utilisation is never stopped for a tip change on schedule. One machine too many is the cheaper mistake, and it is still a mistake: capital sits idle, the operator is underloaded, and the line is designed around a buffer that will be removed in the next cost round.
- Measure seconds per weld on a real machine with a real part, not from a catalogue.
- Use 80 percent availability, not 100, and write the figure down.
- Check the duty cycle at the planned welding-time fraction, not at the average.
- Decide what a bad day looks like before deciding how many machines cover it.
Capacity is not a machine specification. It is a decision about how much slack the plant is willing to pay for, and it belongs in the enquiry rather than in the argument that follows the first missed shipment. Put it in numbers, in writing, before the first quote is compared; the answer is much easier to defend than the opinion it replaces.
PDKJ continues to provide customized laser welding and resistance welding solutions based on different materials, workpiece structures, welding requirements, and production applications.
Related Products
Part of the assembly floor in Dongguan, where a capacity calculation usually gets its first reality check.
Contact PDKJ
Send us your part drawing or a sample, and our engineers will run a welding feasibility trial on your own material and recommend the equipment configured for your application.
WhatsApp: +86 136 1279 3248
Email: jay@gd-pw.com
Website: https://www.spotlaserwelding.com/contactus.html
Address: 1-2F, Building 3, Qichen Industrial Park, No. 26 Luxi 1st Road, Liaobu Town, Dongguan City, Guangdong Province, China