Rework Rate

Goal of the analysis:

Measure the proportion of produced units that require rework or repair before they can be accepted as good, diagnose root causes, and eliminate the “hidden factory” that consumes capacity, extends lead time, and raises Cost of Poor Quality (COPQ). Rework Rate complements First Pass Yield (FPY) and Scrap Rate: whereas FPY focuses on defect-free flow, Rework Rate quantifies units that fail initially but can be recovered. Executives use this analysis to protect margin and service by reducing rework loops at bottlenecks, prioritizing quality-at-source, stabilizing setups/centerlines, improving supplier quality, and targeting error-proofing and training where they pay back fastest.

Data required:

• Production and quality records (MES/ERP/QMS):
  • Units started/completed by operation/line/shift; lot or serial IDs; good vs nonconforming counts.
  • Rework/repair tickets or rework orders (IDs, timestamps, operations performed, disposition, success/fail), scrap transactions.
  • Inspection/test results (pass/fail), defect codes/categories, detection points (in-process, end-of-line, customer returns).
• Routings, standards, and costs:
  • Standard cycle times and rework routings/standard hours (if defined); resource requirements for rework.
  • Labor/overhead rates and material usage for typical rework; warranty/return costs for external rework.
• Process capability and equipment context:
  • SPC data on CTQs (Cp/Cpk, control charts), centerline/recipe parameters, changeover matrices.
  • Maintenance logs (MTBF/MTTR), calibration status, micro-stop history; environmental conditions (temperature/humidity) where relevant.
• Materials and supplier quality:
  • Incoming inspection results, supplier lot IDs/PPM, CoA/COC, alternates/substitutions, containment actions.
• Measurement integrity and staffing:
  • MSA (Gage R&R), false-fail/false-pass rates; operator skill matrix, certification status, training records, shift rosters.
• Policy and normalization:
  • Definition of rework vs normal finishing and cosmetic touch-up; startup window policy; partial acceptance/downgrade rules; time zones/calendars, NPIs/ECNs flags.

Detailed step-by-step instruction on how to conduct the analysis:

1. Define metrics, scope, and counting rules.
   • Rework Rate (%) = Units that required rework (≥1 rework step before acceptance) ÷ Total units produced × 100.
   • Rework Incidence (per 1,000 units) = Reworked units ÷ Units produced × 1,000.
   • Rework Loop Count = Avg number of rework cycles per reworked unit; Rework Success Rate = % of reworked units that become good.
   • Rework Cost ($) = Rework labor + materials + re-inspection + overhead (optionally include warranty for external rework).
   • Capacity impact (constraint days) = Rework hours at the constraint ÷ Effective daily capacity.
   • Clarify inclusions: internal rework only vs include field returns; exclude purely cosmetic “normal finishing” if policy defines it as value-added.
2. Assemble and reconcile data.
   • Extract 8–12 weeks of production, inspection, rework, and scrap data; join by unit/lot IDs and operations; ensure each rework is linked to an originating defect/ticket.
   • Align to common time base and resource IDs; reconcile totals to inventory movements and payroll (for labor hours) to confirm completeness.
   • Validate coding (rework vs scrap vs finishing) via spot checks; correct obvious misclassifications.
3. Compute headline metrics by dimension.
   • Calculate Rework Rate, Incidence, Loop Count, and Rework Cost by plant, line/work center, operation/station, SKU/SKU family, and shift/crew.
   • Overlay FPY and Scrap Rate to show the full quality picture (e.g., low scrap but high rework = hidden factory).
4. Build defect and rework paretos.
   • Pareto rework by defect category (dimension, cosmetic, functional, contamination, assembly error), by operation/station, by supplier lot, and by changeover pair.
   • Map occurrence vs detection points to find where defects are introduced and caught; identify late detections driving rework complexity.
5. Link to process capability and operating conditions.
   • Correlate rework spikes with SPC violations, parameter drifts, changeovers, PMs, and environmental excursions.
   • Check MSA for stations with high rework from false fails; quantify impact of test noise on rework.
6. Translate to business impact.
   • Compute COPQ slice from rework: labor/materials, re-inspection, delay-induced expediting/overtime, potential warranty exposure for rework escapes.
   • Estimate lead time inflation and backlog days from rework minutes at the constraint; quantify OTIF risk.
7. Trend and statistical control.
   • Plot weekly Rework Rate by line/station/shift; use p-charts (proportion defective needing rework) and c/u-charts (rework defects per unit) with control limits to distinguish common vs special causes.
   • Annotate NPIs/ECNs, supplier changes, SMED/centerlining, and maintenance events.
8. Scenario and what-if analysis.
   • Model effect of: SMED (startup rework −30–50%), parameter locks/centerlining, supplier containment (incoming PPM −50%), poka‑yoke at top two stations, MSA improvements (false fails −X%), and earlier detection (in-process checks) on Rework Rate and capacity.
   • Estimate ROI and payback from COPQ savings and throughput gains.
9. Integrity checks.
   • Ensure reworked units are counted once per finished unit (don’t inflate denominator with rework passes); keep rework loops separate from first-pass counts.
   • Confirm that “downgraded” product isn’t recorded as good; verify that rework routings and hours are captured consistently.

Format of the output of analysis:

• Executive scorecard: Rework Rate %, Incidence per 1,000 units, Rework Loop Count, Rework Success Rate, Rework Cost $, capacity days lost at constraint, FPY and Scrap overlays, trend vs target.
• Heatmaps: rework rate by line/work center × shift and by SKU family; concentration of rework by defect category and station.
• Pareto charts: top rework-causing defect modes, stations, supplier lots, and changeover pairs.
• Bridges: COPQ breakdown (labor, materials, re-inspection, overhead) and minutes bridge from total time → rework → recovered capacity.
• Control charts: p-/u-charts for rework proportions/defects with event annotations.
• Action tracker: countermeasures, owners, expected rework reduction, capacity recovered, and verified savings.

How to interpret results:

• High Rework Rate with low Scrap: Hidden factory—many units rescued at cost; capacity and lead time suffer. Prioritize error-proofing and upstream detection to prevent rework introduction.
• Rework concentrated post-changeover/startup: Setup/centerlining weakness; implement first-article checks, parameter locks, and SMED to stabilize startup.
• Rework clustered at few stations/crews: Method or training issue; standardize work, certify skills, and fix fixtures/tools.
• Supplier-lot correlation: Incoming quality problem; enact containment, tighten specs/AQL, and drive corrective action (8D/PPAP refresh).
• High rework with good Cp/Cpk but poor MSA: Test/inspection causing false fails; improve gages and limits to avoid unnecessary rework.
• Rising Rework Rate while OEE falls: Instability (equipment/parameters) or material variance; coordinate TPM with SPC and material controls.

Steps a company can take to improve on this measure:

• Quality at source and process control:
  • Deploy SPC with reaction plans on CTQs; lock centerlines after changeovers; use golden batch references and digital parameter verification.
  • Add in-process detection where late detection drives complex rework; design simple go/no-go checks upstream.
• Error-proofing and method engineering:
  • Install poka‑yoke fixtures/sensors for top assembly errors; standardize work with visual instructions; torque/vision checks where applicable.
  • Streamline rework routings and ensure right tools/fixtures to reduce loops while permanent fixes are implemented.
• Changeover and startup stability:
  • Run SMED to externalize/parallelize setup tasks; implement first-article sign-offs and startup audits; sequence by family to reduce risky transitions.
• Supplier quality and materials:
  • Contain suspect lots; require supplier SPC dashboards and CoA/COC compliance; qualify alternates; enhance lot traceability to accelerate isolation.
• Measurement systems and calibration:
  • Conduct MSA (Gage R&R); recalibrate; tune test limits based on capability and risk; eliminate redundant or damaging tests.
• People, training, and governance:
  • Skill matrices and certification on critical stations; buddying for new/agency workers; layered process audits; daily tiered meetings reviewing rework Pareto and actions.
  • Align incentives to FPY and rework reduction at the constraint, not just output volume.
• Planning and policy:
  • Set frozen windows to reduce resequencing-driven setup instability; define clear rework coding standards to ensure data integrity; throttle WIP (CONWIP) to avoid rushing and quality slips.
• Example scenarios:
  • If Rework Rate is 7.5% with 40% concentrated in two post-setup stations, run SMED and parameter locks plus poka‑yoke; target ≤4% in 6–8 weeks and recover 2–3 constraint days/month.
  • If a supplier lot drives 30% of rework on a CTQ, implement containment and revise incoming AQL/PPAP; expect ≥70% reduction in related rework next month.
  • If false fails inflate rework at end-of-line test, run MSA and recalibrate limits; reduce rework by 30–50% without raising escapes (verify with audit sampling).

Benchmark comparisons:

General benchmarks (directional):

• Discrete assembly: Rework Rate commonly 2–6% of units on mature lines; top quartile ≤2–3% with FPY ≥98%.
• Process/continuous industries: Internal rework often ≤1–3% of units; sustained >3% suggests parameter drift or contamination control issues.
• High-mix/NPI ramps: Temporary rework rates of 5–10% are common; should halve within 1–2 quarters as controls mature.
• External rework (returns/warranty): strive for ≤0.5–1.0% of shipped units in mature programs.

Constructing internal benchmarks:

• Track 12–16 weeks of Rework Rate by line/station/SKU family and shift; publish quartiles and set redlines (e.g., alert if line Rework Rate >4% for 2 consecutive weeks).
• Pair rework with FPY, DPU/DPMO, Scrap Rate, OEE, and constraint load to prioritize cells with highest capacity and customer impact.
• Set component targets: startup-related rework ≤20–30% of total after SMED; critical CTQ rework trending down with Cp/Cpk ≥1.33.
• Rebaseline after SMED/TPM, tooling or supplier changes, or major ECNs; codify best-performing cells as standard and audit adherence.