Connector and cable-harness capacity is no longer a single factory headcount number, it is a five-node work-center problem where molding, soldering, assembly, potting, and inspection each have independent cycle time, scrap, and changeover profiles that determine whether a delivery promise holds [S2].
For 2026 planning, the practical ceiling sits well below design nameplate: across 3,000+ tracked machines the median runtime is 32.04%, the weighted average is 54.54%, and "no business/orders" alone consumed 22,700 lost hours, so capacity planning must start from a measured OEE baseline, not a nameplate [S3].
Capacity as Three Layers, Not One Number
Design, effective, and actual capacity give three distinct answers to the same question, and a connector line that conflates them usually overpromises on lead time [S3]. Design capacity is the theoretical maximum with no interruptions; effective capacity deducts planned downtime, changeovers, and breaks; actual capacity further subtracts unplanned downtime, scrap, and quality holds [S3]. The actionable gap for connector sourcing is the effective-to-actual delta, which on industry data is large because quality events average 142.6 minutes per occurrence and median changeover is 44 minutes with 56.6% variability, meaning that the same connector family can run at very different effective cycle times on consecutive batches [S3].
The Five Work-Center Nodes That Drive Connector Output
An integrated connector process chain runs molding, assembly, potting, inspection, and packing as one coordinated workflow to limit inter-station handoff loss [S2]. Each node has measurable control points rather than qualitative targets: pin insertion force, mating-gap verification, molding completeness, lock depth, O-ring compression state, terminal contact resistance, pressure/insulation/withstand checks, cut length, strip length, conductor condition, color/sequence consistency, crimp quality, solder quality, and electrical validation, all feeding a batch archive and release step with abnormal isolation plus CAPA [S2]. For harness-heavy orders the chain extends to strip length, conductor condition, and crimp pull-force checks, which on a 2,000-pin run can each be a constraint on its own [S2]. The work-center framing matters because capacity requirements planning (CRP) in MES systems like IFS calculates load per work center and per labor class, then flags overload before a shop order is released, rather than after a missed shipment [S4].
Workforce, Machine, and Material: Three Plans Must Agree
Capacity planning has three resource layers, and a connector program that plans only one will fail at handoff [S1]. Workforce planning sums available work hours against skill distribution, planned absence, and overtime; a four-operator cell at 40 h each gives 160 h raw, but after breaks, training, and changeover attendance effective output hours are typically well below that [S1]. Machine planning tracks cycle time, OEE, and changeover by part number, the median 44-minute changeover and 56.6% variability figure is the single most useful lever to compress on a connector line [S3]. Material planning aligns housing, terminal, seal, and wire inventory with the released schedule, and CRP-style tools then validate that the master schedule is feasible against all three layers before the order is committed [S4][S5].
Selection: Lead Strategy, Capacity Strategy, and Constraint Map
There are three classical lead strategies, each with a different capacity signature, and a connector program should be matched to one rather than blended by accident [S1]. Make-to-stock suits high-volume, mature connector families with stable demand, and uses finished-goods buffer to absorb OEE variance; make-to-order fits custom pinouts, mixed wire AWGs, and lower-volume harnesses where finished stock is uneconomic; assemble-to-order or configure-to-order sits between them and is the common mode for industrial connector vendors serving switchgear and PLC cabinet builders [S1]. On top of the lead strategy, the capacity strategy choice is lead, lag, or match: lead adds capacity ahead of demand (capex risk, lower TCO per unit), lag adds capacity after demand materializes (higher expedite cost, lost-sale risk), and match tracks demand with incremental adds (lowest risk where forecast accuracy is below ±20%) [S1].
KPIs That Expose Capacity Drift Before the Quarter Ends
Schedule adherence, capacity utilization, and plan stability are the three KPIs most connector plants underuse, and they directly predict on-time delivery in the next 30-60 days [S7]. Schedule adherence compares actual run start/finish against the frozen plan, and is the earliest signal that a work center is slipping; capacity utilization measured against effective capacity (not design) is the second signal, anything persistently above roughly 85% effective utilization usually means changeover windows are being eaten, which feeds the third KPI, plan stability, because every late shift cascades into a reschedule [S7]. For connector production specifically, lot-level first-pass yield on crimp and soldering, plus mating-cycle test pass rate, are the two quality KPIs that gate capacity release, since a quality hold at final inspection effectively removes that batch from available output until CAPA closure [S2][S7].
Where Capacity Planning Goes Wrong on Connector Programs
The most common failure mode is treating the molding press as the bottleneck when, on a harness-heavy SKU mix, the real constraint is the crimping and tinning station plus the final electrical test cell, a classic work-center load misallocation that CRP-style simulation is designed to expose [S4]. A second failure is ignoring changeover variability: a 44-minute median with 56.6% spread means a 12-changeover day can range from roughly 7 to 14 hours of pure changeover time, which alone can swing effective output by 15-20% before any cycle-time issue is even counted [S3]. A third failure is releasing a build plan against design capacity rather than effective capacity, which produces quoted lead times that the pressure sensor and harness cells cannot physically meet on a multi-SKU week. Sourcing teams should ask vendors for effective-capacity-based lead times by SKU family, and for batch-level first-pass-yield data, before signing annual volumes.
How ERP/MES Tools Validate a Connector Plan
Modern capacity planning sits inside the MES rather than in spreadsheets, and the validation logic is the same across major platforms [S4][S6]. The system ingests the master schedule plus all shop order requisitions (manual, KANBAN, MRP, or master-schedule generated), calculates load by work center and labor class, then compares it against defined capacity to flag overload, with backward scheduling and optional overlap, and uses the planning time fence to treat near-term demand as firm [S4]. On the SAP S/4HANA side the same logic runs as part of manufacturing execution, and the practical value is being able to simulate "what if we add 10% to forecast" or "what if we lose the B-shift on the crimping cell" before committing to a customer, and the systems will surface the bottleneck work center graphically so planners can re-sequence, add a shift, or split a lot [S6]. For connector production, the highest-value simulation is usually a changeover-compression scenario, because that lever is consistently the largest single source of recovered effective capacity [S3].
Sourcing and Standards Touchpoints
Capacity claims should be auditable against process control, not just throughput, and connector-specific quality standards (USCAR, IEC 60512, IPC/WHMA-A-620 for harnesses) define the control points that a credible capacity plan must respect rather than treat as optional overhead [S2]. The capacity release logic is typically tied to a batch archive that captures first-pass yield, crimp pull force, contact resistance, and withstand voltage, so any attempt to "buy" faster lead time by skipping inspection will surface immediately in the audit trail and in the next flow meter or sensor cabinet return rate [S2]. For buyers, the practical test is whether the vendor can produce, on request, the last three months of work-center-level OEE and lot-level first-pass yield by part number, and whether their capacity plan distinguishes effective from actual output.
For a process-engineering view of how changeover compression feeds into broader factory flow, this static-pressure molding machine spec map for lighting fixture production walks through the same effective-vs-actual capacity logic on a different work-center stack.