OEE Software for Manufacturing
Availability times performance times quality: what the number actually measures, and what you need in place before it is worth computing.
Overall equipment effectiveness is a single percentage: availability times performance times quality. Availability is run time against planned production time, which excludes planned downtime. Performance compares actual output to the ideal cycle rate. Quality is good pieces as a share of total pieces produced.
Availability, performance, and quality measured from production data.
OEE compresses three unlike kinds of loss into one number, which is both its strength and its trap. Availability asks how much of the planned production time the machine was running. Performance asks whether it ran at the rate the process is capable of. Quality asks how many of those pieces were good. Multiply them and ninety percent on each term comes out at seventy-three percent, which is why the metric is unforgiving and why arguing about it is a pastime. The eighty-five percent world-class benchmark comes from high-volume lines running one product for weeks at a time. A job shop changing over three times a shift books that setup as unavailability and scores badly for doing exactly what it sells. WorkCell computes OEE from the transactions the floor is already posting, which is what makes the three terms arguable against evidence rather than against each other's spreadsheets.
Plants with equipment-constrained production, where a bottleneck machine, a molding cell, or a press line decides how much the business can ship, and lost run time costs more than labor.
The operator writes "waiting on material" on a sheet that ends up in a drawer. Nobody aggregates it, so the largest recurring loss on the bottleneck machine is known anecdotally, argued about in meetings, and never costed.
Availability runs against planned production time, so somebody has to decide whether an unstaffed second shift, a planned changeover, or a scheduled service counts. Two departments using two denominators produce two OEE numbers and a standing meeting.
The eighty-five percent figure came from dedicated lines. Applied to a job shop it rewards long runs and punishes changeovers, which quietly pushes the schedule toward batch sizes the business never wanted to carry.
Without a trustworthy ideal cycle time on the routing, actual against ideal cannot be computed. A machine running at seventy percent of its capability reads as fully utilised, and the single largest loss in many plants never surfaces.
An OEE screen bolted onto a plant with no operation-level capture shows a number assembled from estimates. It moves, people react to it, and none of it traces back to a shift that actually happened on a specific machine. Within a month the floor has learned the screen does not describe their day, and the project is finished whether or not anyone says so.
Quantity and scrap captured at the operation
Operators record good pieces and scrap at the terminal as they run, against the work order operation. That is the quality term of OEE, recorded where it happens rather than reconciled from paperwork at the end of the week.
Run time from labor tickets
Clocking on and off an operation produces a labor ticket against the resource with real start and stop times. Those tickets are the raw material for the availability term, and they are a by-product of running the job.
Standards on the routing to measure against
Each routing operation carries setup and run time for its resource, so actual time has something to be compared with. Without that standard the performance term of OEE cannot be computed at all, by any tool.
Meters on the resource
A resource can carry meters, counter or gauge, in hours, cycles, pieces, or your own unit, with every reading kept: who took it, when, the delta, and resets handled explicitly instead of by overwriting the total.
A signal endpoint you can post to
Each resource can hold a connection with its own token and an endpoint that accepts JSON over HTTPS. A PLC gateway, an MTConnect adapter, or a bridge on the shop network publishes the same way. You supply the hardware.
Data out through the API
Labor tickets, quantities, scrap, and meter readings are reachable over the REST API with webhooks, so a team that wants an OEE board today can build it on records WorkCell already holds rather than on a new collection effort. The three terms are a query over data the floor is entering anyway, which is the part most plants are missing.
Operator terminals capturing quantity, scrap, and time at the operation.
EngineeringRoutings carrying the setup and run standards performance is measured against.
QualityScrap and nonconformance tied to the operation and resource that produced it.
AnalyticsReporting over the same labor tickets and quantities the floor recorded.
MES records the events: clock-ons, quantities, scrap, downtime reasons. OEE is one metric computed from those events. Buying an OEE tool without a system that captures operation-level data gives you a dashboard with estimates behind it.
OEE vs CMMSOEE measures the loss; a CMMS is one of the ways you reduce it. Availability falls when machines break, so OEE is usually the number a maintenance program gets judged by. Neither replaces the other, and the overlap is the asset record.
OEE vs ERPThe ERP measures money and jobs: hours booked, cost absorbed, orders shipped. OEE measures the equipment. A job can look profitable on a machine running at half its capability, and that is precisely the loss job costing hides.
OEE vs APSOEE tells you how much capacity you lost last week. APS decides how to spend the capacity you have next week. The link between them is efficiency: a scheduler planning a resource at full rate when it runs at seventy percent will always be late.
Molding cells where cycle time and changeover decide how much the plant ships.
Metal ManufacturingStamping and machining where one bottleneck machine sets total plant capacity.
Food & BeveragePackaging lines where availability loss, not labor, is the dominant constraint.
AutomotiveTakt-driven lines where a minute of unplanned downtime has a known cost.
How is OEE calculated?
Multiply three ratios. Availability is run time divided by planned production time. Performance is ideal cycle time times total pieces, divided by run time. Quality is good pieces divided by total pieces. Each is a percentage, and because they multiply, the combined figure falls fast.
Does WorkCell give us OEE?
Yes, per machine and per work centre, built from run time on the labor tickets, good and scrap quantities at the operation, the setup and run standards on the routing, and the meter readings on the resource. Because every term traces back to the transaction behind it, a number nobody believes can be opened up to the shift and the job that produced it, and the same data moves out over the API if you want it on a plant board.
What counts as planned production time?
Planned production time is scheduled hours minus planned downtime: unstaffed shifts, planned maintenance, and periods with no demand. Changeovers normally stay inside it and count against availability. Decide the convention once and write it down, because changing it later rewrites every historical comparison you have.
Is 85 percent OEE a realistic target for a job shop?
No. That figure comes from high-volume discrete lines running one product for long stretches. A high-mix shop changing over several times a shift books setup as unavailability, so a healthy and profitable job shop can score in the forties and be running perfectly well.
Can WorkCell read signals directly off our machines?
Yes. A resource holds its own connection with a token and an endpoint that accepts JSON over HTTPS, so a PLC gateway, an MTConnect adapter, an OPC UA client or an MQTT bridge publishes cycles, counts and meter readings straight onto the machine record. The signal lands on the same resource the routing schedules and the operator clocks against, which is what keeps an automatic count and a human one from becoming two versions of the day.
Should we measure OEE on every machine?
Measure it on the constraint. OEE on a machine that is not the bottleneck improves a number without adding capacity to the plant, and the collection effort is real work for the floor. One bottleneck measured honestly beats forty measured by estimate.
One system instead of a standalone OEE.
Tell us which machine constrains your plant and we will show you every number WorkCell records against it, and how its OEE is built from them.