OEE Lab / Tools / Machines per Operator Calculator

Machines per Operator Calculator, with Machine Interference

How many identical machines can one person tend? Enter the load time, the walk time and the machine's unattended run time. See the ideal number, who waits for whom in each cycle, the output per hour and which whole number of machines gives the lowest cost per unit. Updates as you type.

One operator, identical machines

The defaults are an example cell, not a benchmark. Time the real cycle on your floor and replace them.

Times per machine, per cycle
All three times below use this unit.
Unload, load, quick check and restart. The machine stands still while this happens.
The automatic part of the cycle, when the machine runs without the operator.
Walking to the next machine, plus work done while the machines run: deburring, packing, gauging a part.
Assignment
A whole number. The result compares it with the numbers around the ideal.
Costs per hour (optional)
Wage plus employer costs.
Without labour. Build it with the machine hour rate calculator.
Defaults: 1 + 4.2 min machine cycle, 0.6 min walk
Ideal machines per operator
-
n' = (a + t) ÷ (a + b)
0%operator utilisation
Cycle time
-
Waiting / cycle
-
Operator utilisation
-
Machine utilisation
-
Output / hour
-
Cost / unit
-

One cycle: who waits for whom

Compare machines per operator

The cheapest whole number is always the whole number just below n' or the one just above it.

The working

    This is the deterministic synchronous servicing model. It assumes every cycle is the same and no machine stops on its own. Real breakdowns, jams and slow loads make machines wait for each other at random, and that extra interference grows with each machine you add.

    Treat n' as an upper limit. Check the assignment on the floor for a few shifts before you change the manning.

    Get a second opinion on this result

    Send it to the OEE Lab team. A Fabrico OEE specialist reads your numbers and gets back to you by email. To keep the result, use Copy link or Print above.

    Sent. A Fabrico OEE specialist will read your numbers and get back to you by email.

    No newsletter, no spam. What you enter here, this result and its link go to the OEE Lab team. Privacy

    The formula assumes no machine ever stops on its own.

    Your floor does not. Fabrico records stops live from the PLC or IoT sensors, and the operator gives the cause, such as waiting for an operator. Where a machine has no usable signal, AI cameras count the output. You see OEE per line and per shift, so you can tell whether a manning change paid off or only moved the waiting around.

    See where your machines really wait: book a demo with a Fabrico consultant.

    Request a demo
    Our tools stay free either way.

    What the machines per operator calculator does

    Many cells run several automatic machines with one person: CNC lathes, injection moulding presses, grinders, winding machines. The operator unloads and loads one machine, walks to the next, and so on, while the others run on their own. Work study calls this multi-machine manning or machine tending.

    The question is how many machines to give one person. Too few, and the operator stands waiting for a machine to finish. Too many, and a finished machine stands waiting for the operator. That second wait is called machine interference. This calculator finds the balance point, shows both kinds of waiting for the number you choose, and compares the cost per unit of the whole numbers around it.

    It looks at one operator and one group of identical machines. For the bigger picture of how well your people's time turns into good output, Fabrico has a guide to overall labor effectiveness (OLE).

    The formulas

    Three times per machine, per cycle:

    • a = operator time at the stopped machine: unload, load, quick check, restart. The machine cannot run during it.
    • t = machine run time: the automatic part of the cycle, with no operator needed.
    • b = walk and other work per machine: walking to the next machine, plus any work the operator does while the machines run.

    One machine needs a + t for a full cycle. The operator needs a + b for each machine visit. The ideal number of machines is where the two match:

    n' = (a + t) ÷ (a + b)

    For a whole number of machines n, the cycle is set by whichever is slower, the machine or the operator's round:

    Cycle time T = max(a + t, n × (a + b))

    • If n is at or below n', the machines set the pace. T = a + t. The operator is idle for (a + t) - n × (a + b) each cycle.
    • If n is above n', the operator sets the pace. T = n × (a + b). Each machine waits n × (a + b) - (a + t) for the operator. That is machine interference.
    • Operator utilisation = n × (a + b) ÷ T.
    • Machine utilisation = (a + t) ÷ T: the share of the cycle the machine is being served or running, not waiting.
    • Output per hour = n × 60 ÷ T with T in minutes, one unit per machine cycle.

    Cost per unit = (operator cost per hour + n × machine cost per hour) ÷ output per hour

    Below n', adding a machine adds output at the same cycle time, so the cost per unit falls. Above n', the cycle stretches with every machine added, so the cost per unit rises. The cheapest whole number is therefore the one just below n' or the one just above it, and the table compares both.

    Worked example: the default numbers

    The defaults are an example, not a benchmark. An operator spends a = 1.0 min unloading, loading and restarting each machine. The machine then runs t = 4.2 min on its own. Walking to the next machine and gauging a part take b = 0.6 min. The operator costs 32 an hour and each machine 20 an hour, without labour.

    The ideal is n' = (1.0 + 4.2) ÷ (1.0 + 0.6) = 5.2 ÷ 1.6 = 3.25 machines. So the choice is 3 or 4.

    Stepn = 3n = 4
    Operator round, n × (a + b)3 × 1.6 = 4.8 min4 × 1.6 = 6.4 min
    Cycle time Tmax(5.2, 4.8) = 5.2 minmax(5.2, 6.4) = 6.4 min
    Who waitsoperator, 0.4 mineach machine, 1.2 min
    Operator utilisation4.8 ÷ 5.2 = 92.3%6.4 ÷ 6.4 = 100%
    Machine utilisation5.2 ÷ 5.2 = 100%5.2 ÷ 6.4 = 81.3%
    Output per hour3 × 60 ÷ 5.2 = 34.64 × 60 ÷ 6.4 = 37.5
    Cost per hour32 + 3 × 20 = 9232 + 4 × 20 = 112
    Cost per unit92 ÷ 34.6 = 2.66112 ÷ 37.5 = 2.99
    Both columns use a = 1.0, t = 4.2 and b = 0.6 min. Values are rounded.

    The fourth machine adds 8.3% more output, but each unit costs 12.4% more, because four machines now wait 1.2 min every cycle. With these costs, 3 machines per operator is cheaper. If the operator cost much more than the machines, the answer could flip to 4: try it with the operator at 60 and the machines at 5.

    A second check: a = 1 min, t = 4 min, b = 0.5 min, operator 12 an hour and machines 18 an hour. Then n' = 5 ÷ 1.5 = 3.33. Three machines give a 5 min cycle and (12 + 54) ÷ 36 = 1.83 per unit. Four give a 6 min cycle and (12 + 72) ÷ 40 = 2.10 per unit. Three wins.

    What machine interference does to OEE

    A machine that waits for its operator is in planned production time. The wait is not planned downtime, so it is a loss inside OEE. How it shows up depends on how you record it:

    • Logged as a stop, for example with a reason like waiting for operator, it lowers availability.
    • Left inside the cycle, because the machine's stop is too short to count, it shows up as a longer cycle and lowers performance against the ideal cycle time.

    When the ideal cycle time is the machine's own cycle, a + t, the loss is the same amount either way. In the example with 4 machines, the machine needs 5.2 min but cycles every 6.4 min. Even with no breakdowns, no slow running and no scrap, its OEE could not pass 5.2 ÷ 6.4 = 81.3%. Assigning 3 machines removes that loss and moves the waiting to the operator, which OEE does not see. That is why a manning decision belongs next to the OEE numbers, not only the labour budget.

    Random stops make it worse. When one machine jams, the operator spends time there and the other machines finish and wait. The model above leaves this out. Queueing models of machine interference (the machine repairman problem) handle random stops, but they need data on how often and how long machines stop, which is what a live stop record gives you.

    How to use it on the floor

    1. Time the cycle, not the standard. Watch at least ten cycles on each machine and split them into a, t and b. The cycle time study calculator helps with the averages.
    2. Sort each task into a or b. If the machine must stand still for it, it is a. If the operator can do it while the machine runs, it is b.
    3. Measure the walk on the real layout. Machines placed in a U or face to face shorten b and raise n'.
    4. Enter both costs. Use a machine rate without labour, so the operator is not counted twice.
    5. Run the assignment for a few shifts and record the waits. If machines often wait for the operator beyond what the model shows, random stops are adding interference. Step down one machine, or attack the stops first.

    Common mistakes

    • Putting internal work into a. Deburring or packing a part while the machine runs belongs in b. Putting it in a makes the machine cycle look longer than it is.
    • Rounding n' up by habit. More machines per person is not always cheaper. When machines cost more per hour than people, the extra machine mostly buys waiting.
    • Ignoring the walk. Half a minute between machines looks small, but it is paid on every machine in every cycle.
    • Using the formula for different machines. It assumes identical cycles. Mixed machines need a multiple activity chart with each machine's own times.
    • Leaving out breaks and relief. When the operator leaves for a break, every machine in the group stops after its cycle. Plan relief, or the output per hour will not hold over a shift.
    • Not counting the wait in OEE. If the operator's round sets the pace, the machines lose time every cycle. Record it, or the OEE looks better than the cell runs.

    Machines per operator FAQ

    How many machines can one operator run?

    For identical machines, n' = (a + t) ÷ (a + b), where a is the operator's time at the stopped machine (load, unload, inspect, restart), t is the machine's unattended run time and b is the walk and other work per machine visit. With a = 1 min, t = 4.2 min and b = 0.6 min, n' = 5.2 ÷ 1.6 = 3.25 machines.

    What is machine interference?

    It is the time a machine stands idle, ready for service, because its operator is busy at another machine. It appears when one person is given more machines than n'. The machine is planned to run, so the wait is a loss inside OEE's planned production time.

    Should I round n' up or down?

    Compare both. Rounding down keeps every machine running but leaves the operator idle part of each cycle. Rounding up keeps the operator busy but makes the machines wait. Which costs less per unit depends on the operator's cost per hour against the machine's. The calculator works out both.

    Does waiting for the operator count in OEE?

    Yes. The machine is in planned production time while it waits. If you log the wait as a stop, it lowers availability. If you leave it inside a longer cycle, it lowers performance against the ideal cycle time. When the ideal cycle time is the machine's own cycle, a + t, OEE drops by the same amount either way.

    Why is the real number of machines per operator lower than the formula?

    The formula assumes every cycle is the same and no machine stops on its own. Breakdowns, jams, material problems and variation in load times make machines wait on each other at random, and that grows with each machine you add. Use n' as an upper limit, then check it on the floor.

    Where do I get the machine cost per hour?

    Use your costing system's machine rate, or build one from depreciation, interest, floor space, power, tooling and maintenance with the free machine hour rate calculator on OEE Lab. Leave the operator out of that rate, because this calculator adds the operator separately.

    ← All OEE Lab tools · Open the OEE calculator