Defining Time-to-First-Run
Robot evaluations often center on throughput: peak parts per hour, cycle time, and uptime percentage. These metrics make sense for stable lines producing one product. They show how the robot performs once setup is complete and operation is running properly.
In variable production, those figures leave out important operational conditions. A system with excellent peak throughput but slow task adaptation can underperform a system with lower peak throughput and quick, dependable task changes. For flexible manufacturing, actual economics depend on the proportion of productive running time left after adaptation overhead.
Two more useful measures for robots in variable production are time-to-first-run and time-to-steady-state. Neither measure has broad standardization, so careful measurement is necessary.
Defining Time-to-First-Run
Time-to-first-run (T1R) is the calendar time between the decision to deploy a robot on a new task and that task's first successful production run. It measures wall-clock time, not work effort. If an integrator appointment requires a three-day wait, those three days are included in T1R, even when no work occurs during the wait.
T1R covers every step: scheduling an integrator or technician, demonstrating or programming the task, running tests, verifying quality, and obtaining sign-off. It indicates the minimum period a plant must schedule around when adding a task to a robot cell's repertoire.
Why use wall-clock time rather than work-effort hours? Production schedules operate on calendar time. When a plant manager asks, "when can the robot start on this new part?" the needed answer is a date, not an engineering-hours estimate. T1R in calendar days can directly support production scheduling. Work-effort hours need assumptions about resource availability before they become a date, and those assumptions introduce most estimation error.
Defining Time-to-Steady-State
Time-to-steady-state (TSS) measures the elapsed time from the first production run to the point when the system reaches its target success rate at target throughput. The first run is generally not steady state. Success rates are lower, cycle times can be longer while the system or operator compensates for early performance, and manual intervention is expected more often.
TSS is important because ramp-up economics differ from steady-state economics. During ramp-up, intervention labor offsets part of the automation benefit. A production order begun before steady state may end before steady state is reached, so that batch never receives the full automation benefit.
For demonstration-learning systems, TSS usually means the additional demonstration sessions required after first deployment to reach target policy performance. Additional demonstrations improve the policy by covering more of the part variation the system will encounter. TSS ends once the improvement rate flattens and performance falls within the target operating envelope.
How to Measure These in Practice
To measure T1R, define a start trigger and an end trigger, then record the timestamp for each.
A clear start trigger is the formal recording of the decision to add a new task in the production system. With an MES or work order system, creating a work order for the new task is a reasonable starting point. The end trigger is sign-off on the first production run, when the task passes quality verification and is cleared for normal production.
If the facility has not tracked this historically, approximate T1R can be reconstructed from MES work order creation timestamps and first production event timestamps. This estimate may be imprecise, especially when work order creation does not accurately show when the decision was made, but it provides a useful baseline.
Measuring TSS requires an explicit target success rate, which is often the more difficult step. "Good enough to run" is not measurable. A defined target, such as 96 percent single-attempt success at target cycle time with no manual intervention required, is measurable. After the target is defined, TSS is the number of production shifts, or demonstration sessions in a learning system, required to meet and sustain it for at least two consecutive measurement periods.
Comparing Across Systems
For variable production, T1R and TSS enable direct comparisons between robot systems that throughput metrics alone cannot support.
Take a traditional scripted programming system and a demonstration-learning system. The scripted system may deliver a lower steady-state cycle time for one task because its programming can be precisely optimized for that task's exact geometry. The demonstration system may have a higher steady-state cycle time, yet reach steady state in hours instead of weeks and complete its first run in a day instead of a month.
For 500 units produced over two shifts, the shorter T1R and TSS can more than offset the higher cycle time. Total cost, including setup time and the opportunity cost of the robot cell during the T1R and TSS phases, may strongly favor the faster-to-deploy option despite lower peak throughput.
The crossover point varies with batch size, task change frequency, and the throughput gap between the approaches. No universal answer exists. Measuring these values and comparing them with actual batch size distributions provides a grounded basis for the decision.
What These Metrics Do Not Capture
T1R and TSS do not show variation in task difficulty. Some tasks take less time to set up than others, regardless of programming approach. A system that performs well on easy tasks but struggles with difficult ones should be assessed across the plant's actual distribution of task types, rather than through selected examples.
These measures also omit the cost of failure modes. A system with fast T1R but frequent crashes during TSS may cost more than its metrics indicate, since crashes cause damage, require safety checks, and restart the TSS clock. Proper measurement therefore requires tracking both average performance and the tails of the performance distribution.
When applied carefully, T1R and TSS give plant managers a direct comparison against what matters in variable production: how quickly the cell reaches target performance and the total overhead required to get there. These questions have clearer economic meaning than peak throughput and are worth supporting with measurement infrastructure.