Overlapping the Stations: When Sequential Cycle Time Becomes the Bottleneck
A pick-and-place cell. The robot picks from the feeder, moves 300 mm, places at the station, waits for the operation to finish, then returns. The cycle was 12 seconds. The customer needed 8. We didn’t buy a faster robot. We overlapped the stations.
The original sequence
The robot did everything in order: pick, move, place, wait for press, retract, move back, pick. The press took 4 seconds. The robot waited. The robot’s motion was 6 seconds. The press operation was 4 seconds. Total: 10 seconds of motion plus wait, plus 2 seconds of gripper open/close = 12 seconds. The robot was idle for 4 seconds every cycle while the press did its work.
What “parallelize” actually means
Parallelization means the robot doesn’t wait. While the press is doing its 4-second operation, the robot is already back at the feeder picking the next part. The press cycle and the robot’s return happen at the same time. You need two fixtures: one at the press (where the current part is being worked on), and one at the pick position (where the next part waits).
With two fixtures:
- Robot places part at press fixture (1.5 s)
- Press starts cycling (4 s) and robot immediately leaves
- Robot moves back to feeder (1 s)
- Robot picks next part (1 s)
- Robot moves to press (1.5 s)
- Press finishes and robot swaps the part (1.5 s)
The robot’s 3.5 seconds of return, pick, and move happens while the press runs its 4-second cycle. The robot finishes picking and is back at the press just as the press opens. The press never waits. The cycle time drops to about 7 seconds. We needed 8. Done.
The fixture layout that makes this work
Two positions on the robot: a waiting fixture and a working fixture. The robot arm has two grippers on the same tool, one empty and one holding a part. When the robot arrives at the press, it swaps: the finished part comes out, the new part goes in, in one motion. The press opens, the arm rotates, the press closes. No move back and forth cycle.
| Approach | Cycle time | Robot idle time | Hardware needed |
|---|---|---|---|
| Single fixture, sequential | 12 s | 4 s/cycle | Standard |
| Two fixtures, dual gripper | 7 s | 0.5 s/cycle | Dual-gripper tool |
| Twin robots | 6 s | 0 | Second robot and cell space |
| Faster robot | 10 s | 4 s | New robot at high cost |
The calculation
The bottleneck time is the longer of: the operation time, or the robot’s round-trip time between stations. For parallelization to work, the robot’s return path must fit inside the operation time.
t_cycle = max(t_operation, t_robot_roundtrip) + t_swap
For t_operation = 4 s, t_robot_roundtrip = 3.5 s, t_swap = 1.5 s. The theoretical is close to 5.5 s, but the press open and close adds 1.5 s that the robot waits at the start. The formula gets you close; the layout verification gets you the real number.
When parallelization doesn’t help
If the operation is fast (1 second) and the robot move is 6 seconds, parallelizing doesn’t help. The robot is still the bottleneck. You need a faster move (shorter distance, lighter payload, higher acceleration) or a simpler robot path. Don’t buy dual grippers for a case where the operation is already faster than the robot.
If the operation requires the part to be held by the robot (dispensing, laser marking), you can’t parallelize. The robot has to hold the part during the operation. In that case, look for two stations so the robot dispenses on one while the other cures.
The diagnostic: Time the cycle. Find where the robot is waiting. If it’s waiting for a process (press, curing, heating), overlap the return trip. If it’s waiting for itself (slow move, heavy part), don’t parallelize. Reduce the move time instead.
