# Robot cycle time calculation: worked example > Break a robot cycle into motion, process, gripper and handshake time, with a worked example, availability factor and ways to cut cycle time. Canonical: https://roboticsengineering.in/robot-cycle-time-calculation/ | Published: 2026-10-28 | Updated: 2026-10-28 | Author: Jignesh Panchal **Most robot cycle time estimates are wrong in the same direction: too optimistic. Engineers add up the fastest motion of each axis and forget acceleration, handshakes and the two seconds the gripper spends waiting for a signal.** This guide shows how to build a cycle time from the actual elements that consume it, how to sanity-check the answer before you buy, and what to change when the measured cycle misses the target. ## What counts as one cycle A cycle is the repeatable unit of work the cell must complete: pick a part, place it, return to ready. Everything between two identical states of the robot and the line counts. If the cell makes two different parts, it has two cycles, and the slower one governs the takt. Define the start and end points physically, not by program line. A common definition is “robot at home position with the previous part released and the next part available”. Write it down, because buyer and integrator often assume different definitions and discover the mismatch during acceptance testing. ## The five elements of a cycle | Element | What it includes | Typical share | | --- | --- | --- | | Transfer motion | Robot moving between pick and place points | 35–55% | | Process time | Welding, dispensing, screwing, scanning, pressing | 10–45% | | Gripper action | Open, close, vacuum build, blow-off | 5–15% | | Handshake and wait | PLC signals, safety gates, conveyor index, vision trigger | 10–25% | | Dwell and settle | Vibration settling, part seating, sensor confirmation | 3–10% | *Shares vary enormously by application, but the point holds: motion is rarely the whole cycle.* ## Estimating transfer time honestly Robot datasheets quote maximum TCP speed. That number is almost never reached on a short move, because the arm has to accelerate, travel and decelerate. For moves shorter than roughly one second, the arm spends most of the move accelerating or braking and never reaches rated speed. A workable rule for a first estimate is to take the straight-line distance between points, divide by an effective average speed of 40 to 60 percent of the rated maximum, and then add 0.15 to 0.25 seconds for acceleration and settling. For a 20 kg payload robot rated at 2 m/s, a 1.2 m move typically lands between 1.1 and 1.5 seconds in real production, not the 0.6 seconds a naive division suggests. ``` Example: one pick-and-place cycle Move to pick point 1.30 s Gripper close and confirm 0.35 s Move to place point 1.45 s Gripper open and blow-off 0.40 s PLC handshake and conveyor 0.50 s Return to ready 0.90 s Settle allowance 0.10 s ------------------------------------ Total 5.00 s -> 720 cycles/hour theoretical Apply 85% availability 4.0 s effective -> 612 cycles/hour ``` The availability factor is not padding. Real cells stop for part jams, upstream starvation, quality holds and operator intervention. For a well-run single-part cell, 85 to 90 percent is realistic; for a cell handling several part variants, 75 to 85 percent is more honest. ## Process time dominates in welding and dispensing In a handling cell, motion dominates. In welding, dispensing or scanning, process time often does. A weld that requires 22 seconds of arc-on time cannot be compressed by a faster robot; the only levers are travel speed within the weld, fewer passes, or splitting the work across two stations. For these cells, calculate process time first and treat motion as the gap between processes. A two-station layout that lets the robot weld at one fixture while the operator unloads the other typically recovers 25 to 40 percent of the cycle compared with a single-station cell, because the manual work happens during arc time rather than between cycles. ## Handshakes are where estimates die Every device in the cell talks to the robot through the PLC or directly through I/O. Each exchange costs a scan cycle plus the physical action. Individually these are milliseconds; added up across a cycle they are often half a second to a full second. - Conveyor index and confirmation: 0.2–0.6 s. - Safety gate or light curtain reset: 0.3–1.5 s depending on configuration. - Vision trigger, exposure, processing and result: 0.05–0.5 s. - Servo gripper close plus confirmation: 0.15–0.4 s. - Fixture clamp and pressure switch confirmation: 0.3–0.8 s. None of these is a fault. They are the price of a cell that knows what it is doing. The error is leaving them out of the estimate and then blaming the robot when the cell runs slower than the spreadsheet. The control architecture that governs them is covered in the [robot cell control guide](https://roboticsengineering.in/plc-scada-robot-cell-india/). ## Cycle time versus takt time Takt time is the rhythm the market demands: available production time divided by required output. A single-shift plant running 450 minutes and needing 900 parts a day has a takt of 30 seconds per part. Cycle time is what the cell actually delivers. Design the cell to run at 80 to 85 percent of takt, not at 100 percent. A cell running at exactly takt has no capacity to absorb a minor stop, and the first jam becomes a missed shift. The same discipline applies when the cell is modelled in the [ROI calculation](https://roboticsengineering.in/robot-roi-calculation/), where over-optimistic cycle times are the most common reason a project looks better on paper than in the plant. ## How to measure a cycle you already have Three methods, in increasing order of accuracy. Use at least two of them before quoting a number to management. - **Stopwatch over 30 cycles.** Cheap and surprisingly good, provided you time the same definition of start and end every time and discard the first five cycles while the operator settles into rhythm. - **Robot controller log.** Most controllers record program execution time and I/O transitions. Export a shift and look at the distribution, not the average; the 90th percentile is what the line actually feels. - **PLC or OPC timestamping.** The most useful because it separates robot time from wait time. If the robot is idle 30 percent of the cycle waiting for a conveyor, the robot is not the constraint and buying a faster arm will not help. ## Reducing cycle time without buying a bigger robot In rough order of return on effort: - Remove waits. Parallelise the gripper action with the last part of the move, and trigger vision before the robot arrives rather than after. - Shorten the path. Fly-by points and rounded corners cut travel time more than higher speed does, because the arm never has to stop. - Pre-position. Keep the robot near the pick point at the end of the cycle instead of returning to a distant home. - Reduce tool weight. Every kilogram on the flange costs acceleration across the whole cycle, and a lighter gripper can be worth more than a larger robot. - Split the process. Two fixtures, one robot, alternating, hides manual load and unload time entirely. - Move the work, not the robot. Presenting parts on an indexer or conveyor at the right height and angle shortens every move. Path and motion tuning is done in the robot program, and the cycle-measurement habits that keep it honest are the same ones described in the [teach pendant programming guide](https://roboticsengineering.in/teach-pendant-programming-guide/). If the part position is uncertain, no amount of tuning beats a proper [TCP calibration](https://roboticsengineering.in/robot-tcp-calibration-guide/). ## The estimate that belongs in the quotation Ask every integrator for a cycle time broken into the five elements, with the source of each number stated. An integrator who can show you the motion times from a simulation, the process time from a weld procedure, and the handshake times from the PLC specification is telling you something about how the project will be run. An integrator who quotes a single number with no breakdown is quoting hope. Also ask what cycle time is guaranteed and how it will be verified at acceptance. A guarantee measured on the integrator’s fixture with the operator helping is not the same as a guarantee measured on your production parts during a normal shift. **How do I calculate robot cycle time?** Break the cycle into transfer motion, process time, gripper action, handshakes and settling. Estimate motion at 40–60% of rated speed, add process time, add each I/O exchange, then apply an 85–90% availability factor. **Why is my robot slower than its rated speed?** Datasheet speed is the maximum TCP speed, reached only on long moves. Short moves are dominated by acceleration and braking, so the effective average speed is usually 40–60% of the maximum. **What is a good cycle time for a pick-and-place robot?** A typical single pick-and-place cycle with handshakes is 4–6 seconds, giving 600–900 cycles per hour theoretical and roughly 500–800 effective. Long moves, heavy grippers and vision add time. **How is cycle time different from takt time?** Takt time is the rhythm the market demands, from available time divided by required output. Cycle time is what the cell delivers. Design the cell to run at 80–85% of takt so it can absorb minor stops. **How can I reduce robot cycle time without buying a bigger robot?** Remove waits, shorten paths with fly-by points, pre-position the arm, reduce tool weight, split the process across two fixtures, and present parts closer to the robot. Each can recover 10–30% before any hardware change. In a [die casting cell](https://roboticsengineering.in/die-casting-robot-india/) the robot share of the cycle is bounded by die-open time, not by its own speed. In [spot welding](https://roboticsengineering.in/robot-spot-welding-india/) the robot travels between welds far more than it welds, so travel time dominates the cycle. A cycle time that is not measured cannot be improved, and the [OEE worked example](https://roboticsengineering.in/oee-calculation-manufacturing/) shows how cycle time feeds the performance factor. ## Sources and further reading - [A3 Association for Advancing Automation](https://www.robotics.org/) — robotics integration and cycle time resources. - [ISO 9283](https://www.iso.org/standard/73933.html) — industrial robot performance criteria and test methods. - [ABB Robotics](https://new.abb.com/products/robotics) — robot motion and programming documentation. - [FANUC Europe](https://www.fanuc.eu/) — application notes on cycle time and motion optimisation.