Hours do not divide by the cavity count
Take the order in the form: 25,000 good parts, 2 percent scrap, a 22 second cycle in a single cavity, and 0.6 seconds added for every cavity beyond the first. One cavity needs 25,511 shots at 22 seconds, which is 155.9 hours of machine time. Four cavities needs 6,378 shots, but at a 23.8 second cycle rather than a 22 second one, so it comes to 42.2 hours.
That is a factor of 3.7, not 4. Go to sixteen cavities and the cycle has stretched to 31 seconds, the shots drop to 1,595, and the hours land at 13.7 — a factor of 11.4 against the single cavity, for sixteen times the cavities. Each doubling gives back less than the last, and the whole curve bends further the larger the per cavity penalty is.
The penalty field is the one that matters
Set the extra seconds per cavity to zero and the table turns into the arithmetic everybody does in their head: hours divide cleanly by the count and sixteen cavities is sixteen times faster. Nobody believes that when it is stated out loud, and almost everybody assumes it when comparing tool quotes.
Where the real number sits is not something a calculator can tell you. It comes out of the fill, pack and recovery time for a bigger shot, the extra plate mass to move, and the ejection on a bigger platen. The moulder quoting the tool has an opinion, and it is worth asking for it as a number of seconds rather than as a cycle for each option, because a number of seconds is the thing you can put in this field and test.
Reading the shot and area columns
Both grow in a straight line with the cavity count while the hours flatten out, which is why the useful count is almost never the largest one that fits the schedule. At sixteen cavities the default job is a 380.6 gram shot and 67.2 square inches of projected area. Multiply that area by whatever cavity pressure your moulder works to and you have the clamp force the job asks for, which on many small presses is the thing that decides the count rather than the hours.
The order is worth doing in that direction. Find the counts the press can hold and deliver first, and only then look at which of them gets the order out in time.
What is left out
Tool cost, obviously, and it is the reason the argument happens at all. Also mould changes, colour changes, purge, and the shots it takes to settle a process each time the tool goes in — that last one scales badly with cavitation and is invisible here. The scrap percentage is steady state only.
The hours are also machine hours rather than working hours. A 42 hour job is not five and a bit shifts unless the press runs unattended, and whether it does is a staffing question that changes the answer more than the cycle does.
Questions people ask
How many cavities do I need for an order?
Divide the good parts by the parts an hour each count actually gives, allowing for the cycle getting longer as the tool gets bigger. On the default job — 25,000 parts, 60 hours available, 22 second single cavity cycle and 0.6 seconds added per cavity — three cavities is the lowest count that comes in under the hours, at 54.8. Four comes in at 42.2. Which of those to build is a tooling price question this page does not touch.
Does doubling the cavities halve the cycle time?
No, and it does not halve the machine hours either. The cycle normally gets longer with cavitation because there is more to fill, pack, cool, open and eject, so the hours fall by less than the count rises. On the default figures, sixteen cavities against one is a factor of 11.4 in hours rather than 16. The size of the shortfall is set by the extra seconds per cavity, which is a field because it belongs to the tool and the press.
What does the projected area column tell me?
How the clamp load grows. Projected area is the shadow the parts cast on the parting line, it scales in a straight line with the cavity count, and multiplied by cavity pressure it gives the force trying to open the mould. On the default job it goes from 4.2 square inches at one cavity to 67.2 at sixteen. This page prints the area and nothing more; it makes no claim about what any particular press will hold.
Should I count start up scrap?
Separately, yes. The scrap percentage here is the steady state figure — parts moulded and rejected during a settled run. Getting a process settled takes shots that belong to no order, more of them on a bigger tool, and on a short run they can outweigh everything the table is showing. Add them to the quantity by hand if the order is small enough for it to matter.
Why does the table let me restrict to powers of two?
Because plenty of tools are laid out as balanced runner systems that come in 2, 4, 8 and 16, and an odd count is either awkward to balance or not on offer from the toolmaker. The even option is there for the same reason with more freedom. Leaving it on every count is useful for seeing where the curve actually bends, even when the count you can buy is not on that part of it.