Report: Capacity, Constraints and Setups

1. Capacity and Constraints

The shop's floor consists of 25 machines across ten work centers. The press brakes are the busiest, with 16,165 machine hours in 2025 (36% of total). Machine hours break into operations per day and hours per operation, shown in routing order below.

The press brakes consistently run at the highest utilization of any work center (average of 84% 2023-2025), followed by the robotic weld cell (70%). Across all work centers, average utilization increased from 63% in 2023 to 75% in 2025 as released work grew 18% (4,315 to 5,081 jobs a year; brake standard hours up 19%) across the same set of machines and shifts.

Capacity Constraints

In 2025, the press brakes ran at 91% average utilization and the robotic weld cell at 82%. Against their 97% uptime, the average production capacity buffer was only 6 points for the press brakes and 15 points for the robotic weld cell, the smallest of all work centers. See Appendix Table 1 for additional detail.

With so little buffer, an ordinary heavy week can take the brakes and the weld cell to maximum utilization: they ran at or above 95% utilization in 31% and 37% of weeks in 2025, about twice as often as any other work center. This is explored in more detail in Section 2. See Appendix Table 2 for additional detail.

Throughout 2025, jobs spent 38,589 days waiting in queue at a work center. The press brake station accounted for 44% of those days, which is a function of its consistently high utilization, as well as its volume of operations. There were 5,566 press brake operations in 2025 (20% of total). The robotic weld cell, which also ran at high utilization, accounted for less than 5% of queue days, because only 420 operations (1.5% of total) passed through it.

The press brakes are therefore the shop's primary constraint on two counts: the highest utilization with the least buffer, and 44% of all queue days. The robotic weld cell spends more weeks at maximum utilization (19 against 16) and constrains the enclosure work that crosses it. These stations are examined in more detail throughout this report.

2. Load and Queue Time

Queue time at each work center is driven by one of two factors. At six work centers it is load: queue time is flat at low utilization, rises steadily as utilization increases, and rises sharply once utilization passes ~85%. At the other four work centers it is schedule: queue times are set by daily dispatch lists and do not move with load; these centers rarely operate near maximum utilization.

Across the six work centers with load-driven queue times, we see that their queue times rise sharply when operating near maximum utilization. As shown in Section 1, the press brake operates at these levels 31% of weeks in a year, and robotic weld 37%.

The queueing curve for each of these work centers shows a nonlinear (convex) relationship with a knee: flat and nearly linear at low utilization, then bending sharply upward and rising as utilization approaches 100%. The curve is fitted on rolling 13-week windows rather than single weeks because a week's queue carries the backlog of the weeks before it, so single weeks scatter around the curve (a week at 100% after a quiet month shows a short wait) while the windowed fit explains 45% of the variation at the brakes against 15% for single weeks.

The press brakes and the robotic weld cell are the two work centers above the knee most weeks: in 2025, they were above 85% utilization in 44 (85%) and 29 (56%) weeks. Both curves show an average queue time at 85% utilization (1.25 days for press brake, 2.35 days for robotic weld) that is significantly higher than the mean queue time at lower utilization bands (0.33 for press brake and 0.95 for robotic weld, below 80%). The brakes seldom run light, so their panel has few points below 70% (29 of 138 weeks); the weld cell, one machine on one shift, swings between idle and saturated weeks (20 below 80%, 19 at or above 95%), climbs faster with load (3.7 days on the curve at 90%) and ran above its curve in 2025 (2.87 days measured against 1.87). See Appendix Tables 3, 4 and 5 for additional detail.

The brake center consists of five machines. Precision work and expedited jobs are routed to B1 and B2, so these two machines have the shortest queue times (2.6 and 2.5 days against 3.2 to 3.5 on B3 to B5) at about the same utilization (90% against 91% to 92%). The long brake waits that drive late jobs are concentrated on B3, B4, and B5, where jobs are routed without a priority queue. See Appendix Tables 6 and 7 for additional detail.

Assembly and the manual weld bays sit below the knee most weeks: in 2025 they were above 85% utilization in 18 (35%) and 9 (17%) weeks. Both show the same step as the brakes: at low utilization levels, mean queue time of 0.84 at assembly and 0.72 at manual weld against 2.24 and 1.68 at or above 85% utilization. The curve fits assembly less closely than the brakes (rank correlation 0.80 against 0.89) and the weld bays less closely again (0.62), since six bays pool arrivals until the last few points of load. See Appendix Tables 3, 5, 8 and 9 for additional detail.

The laser and the punch are first operations, so their queue is measured from traveler print to first cut, less any wait for material. The lasers pool three machines over about 83 operations a week and reach high utilization only when releases surge, with queue time increasing from 0.65 days below 80% utilization to 2.5 at or above 90%. The punch, one machine running about 16 operations a week, fills on a few large jobs, and queue time increases from 0.25 at low utilization to 1.91 at or above 85%. See Appendix Tables 3, 5, 10 and 11 for additional detail.

3. Brake Hours: Setups, Standards and the Peak

Sections 1 and 2 established the press brakes as the shop's primary constraint and mapped how their queue time rises with load. This section is aimed at the brake hours themselves: where hours are lost to setups running over standard and how many can be recovered, whether the standard hours the schedule is planned in are right, and where crewed hours fall short. The next report, Discrete Event Simulation: On-time Delivery, takes these measured hours as its inputs.

Utilization is running machine hours over hours available, and the shop controls both sides: how many hours each job takes, and how many hours are crewed. Downtime (9 hours a week at 97% uptime) and run time over standard (9 hours a week at 1.05 of standard) are small. Setups over standard are not: 36 hours a week, which is 10.4 points of brake utilization against the standards the schedule plans with, and about 7.8 points once the small-lot standards are corrected.

Brake setups over standard

Setups over standard on the brakes ran 1,847 hours in 2025, with the median brake setup running at 1.16 of standard and the 90th percentile at 2.05. This overrun is unique to brakes: no other work center exceeds 51 setup hours over standard in the year or a median ratio of 1.00. See Appendix Table 12 for additional detail.

Four overlapping conditions drove most of the overrun. Setups handed over at a shift boundary are 13% of setups and 51% of the overrun. Lots under 25 pieces are 49% of setups and 67% of the overrun, because the setup standards under-plan small lots (median 1.28 against 1.05 on larger lots). First setups of a part by an operator run at 1.29 against 0.99 when the operator has set it up before. Enclosures and chassis run at 1.31 and 1.29 against 1.09 to 1.10 for panels and brackets. See Appendix Tables 13 to 17 for additional detail.

What can be recovered

Three changes recover 13.9 of the 36 hours a week, 4.1 points of brake utilization, taking the brakes from 91% to 87% and queue time on the curve from 2.34 to 1.46 days per operation. Running the top 12 part-operations by overrun at standard releases 8.1 hours a week; the rest of the overrun is spread across 1,943 part-operations. Assigning repeat setups to an operator who has set the part up before releases 4.3: first setups of a repeat part by a new operator ran at 1.58 of standard against 1.31 with prior familiarity. A handover standard, under which the operator who starts a setup finishes it or leaves a written handover at the machine, releases 1.5, the handover's own cost once long setups are set aside. The report Discrete Event Simulation: On-time Delivery carries these 13.9 hours as the setup program. See Appendix Tables 18 to 21 for additional detail.

Planning standards

The schedule and the quote are built in standard hours, and the brake standards understate what the brakes actually run: the 2025 plan carried 13,940 brake hours against 15,967 run, a gap of 2,027 hours, or 11.4 points of utilization. Two corrections bring part of it onto the plan. Raising the setup standard on lots under 25 pieces by 23% adds 483 hours. Refreshing the 525 stale brake standards (those dated at the part's first quote, which run 10% further above standard than refreshed parts) adds 242. Together they close 36% of the gap, and the setup recovery above takes a further 724 hours off the actual. What remains, 578 hours, is the overrun the plan will still not see. See Appendix Tables 22 to 25 for additional detail.

Away from the constraint, the L2 and L3 lasers run at 0.82 of their run standard, so planned laser hours are overstated by 12.9%; setting those standards to measured is an estimating correction (Flow and On-time Delivery, Section 4) rather than a capacity one.

Where crewed hours fall short

Setup recovery adds 13.9 brake hours a week; the year-end peak needs far more. In the six weeks from November 11, 2024 the shop released 361 brake standard hours a week against 284 of crewed capacity, a shortfall of about 77 hours a week that went into the queue. Queue-triggered Saturday shifts covered almost none of it: 2 shifts in six weeks, 28 hours in total. A Saturday brake shift as the shop has worked it supplies 6.8 machine hours, 8% of the weekly gap; one with both B1 and B2 crewed supplies about 15, 17%. Discrete Event Simulation: On-time Delivery tests one every week from November through February. See Appendix Table 26 for additional detail.

The robotic weld cell runs one shift at 82% utilization and is at or above 95% in 19 weeks of 52; its second shift is uncrewed. Enclosures are 78% of its hours, so a second shift would serve the enclosure customers first. The simulation report tests and prices crewing it. See Appendix Table 27 for additional detail.

Measured from the record: the brakes lose 36 hours a week to setups over standard, enough to put them above the knee of their curve on its own; 13.9 of those hours are recoverable by three changes; the small-lot setup standard, 546 stale standards and the L2 and L3 run standards misstate planned hours in known directions; the year-end peak releases about 77 brake hours a week more than the brakes can run; and the robotic weld cell's second shift is uncrewed. Discrete Event Simulation: On-time Delivery takes these as its inputs.

Appendix

Table 1. Utilization and machine uptime by work center

Table 2. Weekly utilization at or above 95% by work center: weeks, queue time, jobs at the work center and share of hours

2025

2025 Q2-Q4

2023 to 2025

Brake queue 90th percentile by quarter, 2025

Table 3. Weekly queue time by utilization band and work center, full weeks 2023 to 2025

Weeks of 2025 at or above 80% and 95% utilization

Table 4. Fitted queue time curve by work center

Table 5. Position of each work center on its curve

Table 6. Utilization and queue time for the five brakes and the robotic weld cell, 2025 and Q2-Q4

Table 7. The brakes and the robotic weld cell: hours and utilization on both bases

Table 8. Queue days by work center and driver of queue time, 2025

Table 9. Assembly: hot weeks, where its work comes from, and the weekly correlations

By year

* 2023 includes the first four weeks of the record, when jobs already in process raise the queue time; from the week of January 30 the 2023 queue time is 1.00 days per operation.

Weeks at or above 95% against the other weeks, 2025

Where assembly's work comes from, 2025

Hours by part family, 2025

Weekly correlations

Table 10. Laser weeks, November and December 2024

Table 11. The lasers: first-operation queue time and material wait by month, and the three machines

By month, 2025

The three machines, 2025

Table 12. Setup against standard by work center

Table 13. Brake setup against standard by period

Table 14. Brake setups handed over at a shift boundary, and by time left in the shift at the start

Table 15. Brake setups by lot-size band

Table 16. Brake setups by the operator's prior setups of the part

Table 17. Brake setups by part family

Table 18. The top 12 setup reduction candidates by overrun hours times work center utilization, 2025

Table 19. Setup reduction candidates, top 20

Table 20. First setups of repeat parts where another operator had set the part up

Table 21. Brake hours released: the handover and the candidate list

Table 22. Run actual against standard by work center and laser

Table 23. Stale-standard parts against refreshed parts

Table 24. Planned operation hours with standards refreshed

Table 25. Standard hours per job with standards refreshed, by routing class

Table 26. Brake standard hours released against crewed brake capacity, the six weeks from November 11, 2024

Table 27. The robotic weld cell: weeks by utilization band, hours by part family, setup and run against standard

Weeks by utilization band, 2025

Hours by part family, 2025

Setup and run against standard, 2025