Інженерія якості виробництва електросамоката як 31-ша engineering axis: manufacturing-process axis — ISO 9001:2015 + IATF 16949:2016 + AIAG APQP + PPAP + SPC + MSA + AIAG-VDA FMEA + 8D + Lean Manufacturing TPS + Six Sigma DMAIC + Poka-yoke

У серії інженерного гайду ми описали акумуляторну батарею з BMS і thermal runaway intro, гальмівну систему, мотор і контролер, підвіску, шини, світло і видимість, раму й вилку, display + HMI, зарядний пристрій SMPS CC/CV, connector + wiring harness, IP-захист, bearingи з ISO 281 L10, стеблину і механізм складання, деку, handgrip + lever + throttle, колесо як assembly, інженерію різьбових з’єднань як joining-axis, термоменеджмент як heat-dissipation axis, EMC/EMI як interference-mitigation axis, кібербезпеку як interconnect-trust axis, NVH як acoustic-vibration-emission axis, функціональну безпеку як safety-integrity axis, інженерію життєвого циклу батареї як sustainability axis, ремонтопридатність як repairability-axis, environmental robustness як environmental-conditioning axis, privacy і захист персональних даних як privacy-preservation axis, інженерію надійності як reliability-prediction meta-axis, software & firmware engineering як SW-process axis та human factors і ергономіку як human-machine fit axis. Ці 30 engineering-axes описали підсистеми, способи з’єднання, теплові й електромагнітні явища, безпеку, sustainability, ремонтопридатність, environmental conditioning, privacy, reliability-engineering, SW-process та human-machine fit. Кожна з них зафіксувала specifikацію (target dimension + tolerance + material property + test limit) — але жодна не описала сам інструментарій того, як ці специфікації переводяться у production-floor реальність на конкретному виробничому майданчику конкретного дня з конкретною партією компонентів і конкретним оператором.

Manufacturing quality engineering — це production-process axis усього e-самоката. Вона надає процесні стандарти (ISO 9001:2015 QMS foundation + IATF 16949:2016 automotive QMS layered overlay), product-development methodology (AIAG APQP 5-phase framework), supplier-part qualification gate (AIAG PPAP 18-element submission + 5 levels), risk-anticipation tool (AIAG-VDA FMEA Handbook 2019 7-step approach з Action Priority заміняє RPN), statistical control of production variation (AIAG SPC 2nd ed. 2005 з 7 control charts + Western Electric / Nelson rules), process capability quantification (Cp/Cpk/Pp/Ppk indices з threshold values), measurement-system capability quantification (AIAG MSA 4th ed. 2010 з Gage R&R + NDC), post-launch problem-solving (Ford TOPS 8D з 8 disciplines + root cause vs escape point distinction), waste-elimination philosophy (Toyota Production System Ohno + Toyoda з Jidoka + JIT + Andon + Kanban + Heijunka + 7+1 muda), statistical defect-rate methodology (Six Sigma Motorola Bill Smith 1986 з 3.4 DPMO + DMAIC + DMADV), і error-prevention pattern (Poka-yoke Shigeo Shingo 1960s з warning + control types).

Це тридцять перша engineering-axis deep-dive у серії гайду — і чотирнадцята cross-cutting infrastructure axis (паралельна до joining DT + heat-dissipation DV + interference-mitigation DX + interconnect-trust DZ + acoustic-vibration-emission EB + safety-integrity ED + sustainability EF + repairability EH + environmental-conditioning EJ + privacy-preservation EL + reliability-prediction EN + SW-process EP + human-machine-fit ER, тепер manufacturing-process ET). Як і reliability + SW + ergonomics, manufacturing-quality axis не має «залізної» реалізації — це methodology, що визначає, який саме компонент кожної з 30 попередніх axes ви тримаєте в руках: ваш конкретний exemplar відповідає design intent, чи бракований; ваш конкретний brake-pad fall within μ-coefficient tolerance band; ваш конкретний battery cell capacity match nameplate ±5%; ваш конкретний motor stator winding torque sit на target ± 3σ.

1. Manufacturing quality ≠ design quality ≠ inspection: окрема axis

Design engineering і manufacturing quality engineering вирішують різні задачі, які часто плутають:

ВимірDesign engineeringManufacturing quality engineering (ET)Inspection / QC
ПитанняЯкою має бути деталь, щоб система працювала?Як систематично виробити деталь, що відповідає design intent?Чи цей конкретний exemplar відповідає spec?
АртефактDrawing + BOM + DFMEA + design verification reportControl plan + PFMEA + SPC chart + PPAP packageInspection report + reject / accept
Стандарт-фундаментISO/IEC ind.-specific design standardsISO 9001:2015 + IATF 16949:2016 + AIAG core toolsISO 2859 / ANSI Z1.4 / MIL-STD-105E sampling
МетрикаPerformance + safety + costCpk + Gage R&R + first-pass yield + DPPMPPM defect + acceptance / rejection
Цикл валідаціїDV (Design Validation) + DVP&RPV (Process Validation) + PPAP + SPC monitoringLot-by-lot AQL sampling
Тригер“Чи витримає frame 100 000 cycles?”“Чи всі 100 frames партії витримують?”“Чи цей конкретний frame пройшов pull-test?”

Design відповідає на «яким має бути»; manufacturing quality відповідає на «як зробити так, щоб усі exemplars відповідали»; inspection відповідає на «чи цей exemplar відповідає». Manufacturing quality є process between design and inspection — і саме вона унеможливлює покладатись на 100% inspection (надто дорого + людський оператор дає false positive ~5% і false negative ~10% навіть на простих attribute checks).

2. ISO 9001:2015 — QMS foundation

ISO 9001:2015 Quality Management Systems — Requirements опублікований у вересні 2015 — це найбільш поширений management-system стандарт у світі (~1 млн сертифікованих організацій станом на 2023). Він задає загальний фундамент QMS, на якому надбудовуються galuzeві варіанти (IATF 16949 automotive, ISO 13485 medical, AS9100 aerospace, ISO/TS 22163 rail).

Annex SL High-Level Structure — 10 clauses (загальна структура для всіх ISO management-system standards з 2015):

  1. Scope — область застосування.
  2. Normative references — нормативні посилання.
  3. Terms and definitions — терміни і визначення (через ISO 9000:2015).
  4. Context of the organization — interested parties, scope визначення.
  5. Leadership — top-management commitment, quality policy, ролі/відповідальність.
  6. Planningrisk-based thinking (clause 6.1), quality objectives.
  7. Support — resources, competence, communication, documented information.
  8. Operation — design + development + production + service provision controls.
  9. Performance evaluation — monitoring + measurement + internal audit + management review.
  10. Improvement — nonconformity + corrective action + continual improvement.

Сім принципів управління якістю (ISO 9000:2015):

  1. Customer focus — meeting customer requirements + striving to exceed expectations.
  2. Leadership — top management establishes unity of purpose.
  3. Engagement of people — competent + empowered + engaged people.
  4. Process approach — activities understood and managed as interrelated processes.
  5. Improvement — ongoing focus on improvement.
  6. Evidence-based decision making — decisions on analysis of data + information.
  7. Relationship management — managing relationships з interested parties (suppliers, customers, regulators).

Ключове ново в 2015 vs 2008:

  • Risk-based thinking (clause 6.1) — обов’язково ідентифікувати risks + opportunities; “preventive action” як окремий розділ більше не існує (бо risk thinking уже інтегрований).
  • Quality manual більше не обов’язковий — organization сама вирішує scope і format.
  • Management representative role removed — leadership responsibility distributed серед top management.
  • Documented information замінив “documents” + “records” — об’єднана концепція.

3. IATF 16949:2016 — automotive QMS layered on ISO 9001

IATF 16949:2016 Quality management system requirements for automotive production and relevant service parts organizations — опублікований 1 жовтня 2016, замінив попередній ISO/TS 16949:2009 з transition deadline 14 вересня 2018 для всіх existing certifications. Розроблений IATF — International Automotive Task Force, формований 1996 з founding OEMs BMW, Daimler (Mercedes-Benz), FCA Italy, FCA US (Chrysler), Ford, GM, PSA (Peugeot-Citroën), Renault, VW + асоціаціями виробників (AIAG для North America, ANFIA Italy, FIEV France, SMMT UK, VDA Germany).

Ключова особливість: IATF 16949 не є standalone standard — він обов’язково використовується у комбінації з ISO 9001:2015 + customer-specific requirements (CSRs) кожного OEM-замовника. Стандарт додає ~140 додаткових automotive-specific requirements поверх ISO 9001:2015.

Ключові додаткові automotive requirements (відсутні у ISO 9001):

  • Corporate responsibility (clause 5.1.1.1) — anti-bribery policy + code of conduct + escalation policy.
  • Product safety (clause 4.4.1.2) — formal product-safety process з identified safety-related characteristics.
  • Embedded software (clause 8.4.2.3.1) — software development assessment methodology (з посиланням на Automotive SPICE — як у SW-process article).
  • Warranty management (clause 10.2.6) — formal warranty-failure analysis process + NTF (no-trouble-found) tracking.
  • Customer-specific requirements (CSRs) — кожен OEM публікує свої додаткові CSRs (BMW, GM, Ford тощо), які виконавець мусить включити у свій QMS.
  • Manufacturing feasibility (clause 8.3.2.3) — formal feasibility study обов’язковий перед contract acceptance.
  • Special characteristics (clause 8.3.3.3) — identifikація + tracking + control plan inclusion критичних product / process characteristics.

Certification scheme:

  • 3-річний certification cycle з re-certification аудитом на 3-річному cycle end.
  • Surveillance audits annually (typically на 1-й і 2-й рік).
  • Site-specific certification — кожен production site сертифікується окремо; corporate HQ + design centers не можуть отримати independent certification (бо без manufacturing process там нема нічого, що сертифікувати).
  • Certification body — обов’язково має IATF-recognized status (видається через регіональний oversight office: AIAG для North America, etc.).

Status quo (2025): IATF 16949 є de-facto entry ticket для будь-якого Tier-1 / Tier-2 automotive supplier. Якщо самокат-виробник постачає компоненти OEM, який має IATF-certified supply chain (наприклад, спільні harness або lighting suppliers між e-bike / e-scooter і e-car industries), той виробник зобов’язаний бути IATF-certified або pass equivalent customer audit (e.g., VDA 6.3 process audit).

4. APQP — Advanced Product Quality Planning

APQP (Advanced Product Quality Planning) — розроблений у пізніх 1980-х представниками Ford + GM + Chrysler + ASQ як єдина methodology product-development. Поточна reference — AIAG APQP Manual, 2nd edition, 2008 (третє видання вийшло пізніше, але 2nd ed. найчастіше цитується).

П’ять фаз APQP:

ФазаНазваКлючові outputs
Phase 1Plan and Define ProgramVoice of Customer (VoC), product reliability/quality goals, preliminary BOM, preliminary process flow
Phase 2Product Design and DevelopmentDesign FMEA (DFMEA), Design Verification Plan (DV), engineering drawings + specifications, prototype build
Phase 3Process Design and DevelopmentProcess FMEA (PFMEA), process flow diagram, control plan (pre-launch), packaging standards, manufacturing process instructions, MSA plan
Phase 4Product and Process ValidationProduction Trial Run (PTR), Measurement System Evaluation, Production Validation Testing, PPAP submission
Phase 5Launch, Feedback, Assessment, and Corrective ActionReduced variation, customer satisfaction, delivery + service, lessons learned

Ключовий outputControl Plan: документ з 23 monitored topics (part name, process step, machine/tool, characteristic, specification, evaluation/measurement technique, sample size + frequency, control method, reaction plan). Control plan існує у трьох версіях: Prototype (для DV), Pre-launch (для PV), Production (для post-launch ongoing).

APQP working logic: спочатку defines product (Phase 1-2), потім defines process (Phase 3), потім validates обидва на real production tooling (Phase 4), потім continuously improves (Phase 5). Виходи кожної фази є entry gates для наступної — gate review є formal milestone.

APQP-mandated documents для e-scooter component (typical example для brake-pad supplier):

  • DFMEA для brake-pad design (friction-coefficient stability vs temperature, wear rate, noise generation).
  • PFMEA для grinding + bonding + curing process (resin mixing ratio, cure-temperature uniformity, surface roughness).
  • Control plan з 8 monitored characteristics (pad thickness ±0.05 mm, friction coefficient μ 0.40 ±0.05 cold + hot, density 2.1 ±0.05 g/cm³, etc.).
  • MSA plan для μ-coefficient measurement (Gage R&R на dynamometer).
  • Initial process study (30 consecutive parts → Pp/Ppk ≥ 1.67 required).
  • PPAP submission (наступна section).

5. PPAP — Production Part Approval Process

PPAP (Production Part Approval Process)AIAG PPAP Manual, 4th edition, 2006 — formal supplier-customer approval gate для кожної production part. PPAP submission обов’язковий перед:

  • Перший випуск нової part до production.
  • Engineering change (geometry, material, tolerance).
  • Manufacturing process change (нова machine, new tooling, new supplier).
  • Sub-supplier change (sub-tier для key components).
  • Tooling repair / replacement (якщо impact на dimensions).
  • Production restart after extended dormancy (typically > 12 months).
  • Customer request (e.g., quality concern triggers re-PPAP).

18-element submission package:

#ElementЗміст
1Design RecordsDrawing з revision level + math data (CAD)
2Engineering Change DocumentsAuthorized engineering changes
3Customer Engineering ApprovalЯкщо CSR вимагає
4Design FMEA (DFMEA)Якщо supplier має design responsibility
5Process Flow DiagramsManufacturing process flow
6Process FMEA (PFMEA)Для кожного manufacturing step
7Control PlanProduction version з reaction plans
8Measurement System Analysis StudiesGage R&R + Bias + Linearity + Stability
9Dimensional ResultsLayout inspection results vs drawing
10Material / Performance Test ResultsSpecs verification
11Initial Process StudiesPp/Ppk на initial production run
12Qualified Laboratory DocumentationLab accreditation (ISO/IEC 17025)
13Appearance Approval Report (AAR)Якщо visible part
14Sample Production PartsSubmitted physical samples
15Master SampleSample retained by supplier as reference
16Checking AidsTemplates, fixtures, gages used for control
17Customer-Specific RequirementsPer OEM CSR list
18Part Submission Warrant (PSW)Cover-sheet sign-off

П’ять submission levels — level визначає, яка частина 18-element package shipped до customer vs retained at supplier:

LevelЗміст
Level 1Тільки PSW (Part Submission Warrant)
Level 2PSW + product samples + limited supporting data
Level 3PSW + product samples + complete supporting data (typical level для більшості OEM)
Level 4PSW + other requirements as defined by customer
Level 5PSW + product samples + complete supporting data available at supplier’s manufacturing location for review (customer comes on-site)

PPAP outcome: customer approves, interim approves (з deviation), або rejects. До approval, supplier cannot ship production parts (тільки production trial run dimensional limited shipments). Interim approval має expiration і вимагає corrective action plan.

6. AIAG-VDA FMEA — 7-step approach

FMEA (Failure Mode and Effects Analysis) має військове коріння — MIL-P-1629 (1949) + later MIL-STD-1629A (1980). Aerospace adopted early (NASA Apollo + Viking + Voyager). Ford застосував у 1977 (після Pinto affair) — це початок automotive PFMEA.

До 2019 паралельно існували AIAG FMEA 4th edition (2008) для North America і VDA Band 4 для Germany — з різницею у severity tables, occurrence scales, і RPN thresholds. Це примушувало global suppliers тримати дві FMEA-системи паралельно.

AIAG-VDA FMEA Handbook 1st edition, June 2019 — спільна harmonization від AIAG + VDA, прийнята всіма major OEMs (GM + Ford + Stellantis + BMW + Mercedes-Benz + VW + більше). Замінила обидві попередні методології одним 7-step process.

7 steps AIAG-VDA:

  1. Planning and Preparation — scope + boundary + team + foundation FMEAs.
  2. Structure Analysis — block diagram / structure tree.
  3. Function Analysis — function decomposition.
  4. Failure Analysis — failure modes + effects + causes (3-tier).
  5. Risk Analysis — Severity (S) + Occurrence (O) + Detection (D) ratings → Action Priority (AP): high / medium / low (замість old RPN = S × O × D).
  6. Optimization — recommended actions для high + medium AP items.
  7. Results Documentation — official FMEA worksheet + management review.

Ключова зміна — Action Priority заміняє RPN:

Старий RPN = S × O × D мав три проблеми: (1) ordinal-scale multiplication математично некоректна (RPN 80 не “вдвічі гірше” від RPN 40); (2) дві різні комбінації могли давати однакове RPN з різним actual risk (S=10, O=2, D=4 → 80 vs S=2, O=10, D=4 → 80 — перший case є safety-critical один за одним viewpoint); (3) RPN threshold (typically 100 or 125) штучний — між RPN 99 і 100 нема real risk difference.

Action Priority використовує lookup table з 1000 combinations (10 S × 10 O × 10 D), і кожна combination окремо mapped to High / Medium / Low на основі expert judgement OEM + AIAG + VDA technical committee. Severity 9-10 з будь-якими O+D дає High AP автоматично (бо safety/regulatory consequence).

Severity ratings (S, 1-10) — appearance < discomfort < degraded function < major function loss < safety hazard з warning < safety hazard без warning < regulatory non-compliance < total loss + injury з warning < total loss + injury без warning.

Occurrence ratings (O, 1-10) — predictive frequency: 1 ≈ “very low” (< 1 in 1 500 000), 10 ≈ “very high” (≥ 1 in 2).

Detection ratings (D, 1-10) — likelihood that current detection control catches failure mode before customer: 1 = “almost certain detection” (e.g., poka-yoke), 10 = “no detection / no control”.

7. SPC — Statistical Process Control

SPC має коріння у роботі Walter A. Shewhart at Bell Laboratories у 1924 — він винайшов control chart як method для distinguishing common-cause variation (natural noise of stable process) від special-cause variation (assignable cause requiring intervention). Shewhart’s Economic Control of Quality of Manufactured Product (1931) — foundational text. W. Edwards Deming масштабував SPC у WWII US industry + post-war Japan (через JUSE — Union of Japanese Scientists and Engineers), де він вплинув на Toyota Production System. Modern reference: AIAG SPC Manual, 2nd edition, 2005.

Common cause vs special cause — fundamental distinction:

  • Common cause — inherent variation stable process; addressed by redesigning process (machine + material + method change), not adjusting individual measurements.
  • Special cause — external disturbance: shift change, raw material lot change, tool wear, environment swing. Addressed by investigation + corrective action at root cause.

Адresing special cause як common cause = over-adjustment (Deming’s “tampering” funnel experiment). Адresing common cause as special cause = chasing noise (kills productivity).

Сім control charts (AIAG SPC Manual):

ChartData typeUse case
X̄-R (X-bar R)Continuous, subgroup size 2-10Most common variables chart
X̄-s (X-bar s)Continuous, subgroup size > 10Standard deviation more accurate than range
Individuals-Moving Range (ImR / I-MR)Continuous, subgroup size 1Slow process; cost prohibits subgrouping
p-chartAttribute, % defective, variable sample sizeFraction non-conforming
np-chartAttribute, # defective, fixed sample sizeNumber non-conforming
c-chartAttribute, # defects per unit, fixed unitDefects in a constant-size unit
u-chartAttribute, defects per unit, variable unitDefects per unit, variable size

Control limits±3σ from process mean (UCL = μ + 3σ, LCL = μ − 3σ; для X̄ chart σ_X̄ = σ/√n). Це statistical limits, не specification limits — два concept differ:

  • Specification limits (LSL / USL) — встановлені design engineering: «параметр має бути у [LSL; USL] для функції».
  • Control limits (LCL / UCL) — обчислені з process data: «process is stable якщо points у [LCL; UCL] without patterns».

Process може бути in control but not capable (stable, але не fits within spec) або capable but not in control (sometimes meets spec, але unpredictably). SPC + capability аналіз працюють у тандемі.

Western Electric Rules + Nelson Rules — pattern-detection rules для signaling special cause навіть якщо individual point inside ±3σ:

  1. 1 point > 3σ from mean (out of limits).
  2. 2 of 3 consecutive points > 2σ on same side.
  3. 4 of 5 consecutive points > 1σ on same side.
  4. 8 consecutive points on same side of mean (run rule).
  5. Trend of 6 consecutive points increasing or decreasing.
  6. 14 consecutive points alternating up-down.
  7. 15 consecutive points within 1σ of mean (stratification — hidden two populations).
  8. 8 consecutive points beyond 1σ (mixture).

Rational subgrouping — fundamental rule: within-subgroup variation must capture only common cause, while between-subgroup variation captures process shifts + special causes. Bad subgrouping (e.g., grouping samples from different shifts) hides process shifts. Good subgrouping (consecutive parts from same shift) preserves separability.

8. Process capability — Cp / Cpk / Pp / Ppk

Process capability quantifies how well process output fits within specification limits. Чотири pivotal indices:

Cp — Process Capability (potential):

$$C_p = \frac{USL - LSL}{6\sigma_{within}}$$

Cp ignores process centering — measures only spread vs spec width. Process can have Cp = 2.0 (excellent potential) while being off-center і producing 100% defective.

Cpk — Process Capability (actual):

$$C_{pk} = \min\left(\frac{USL - \mu}{3\sigma_{within}}, \frac{\mu - LSL}{3\sigma_{within}}\right)$$

Cpk accounts for centering. Cpk < 0 if process mean is outside specification (i.e., > 50% defects predicted).

Pp + Ppk — same formulas, але σ_total (overall standard deviation) instead of σ_within (within-subgroup standard deviation). Conceptual difference:

  • Cp / Cpk = short-term capability (потенціал з stable process): only within-subgroup variation.
  • Pp / Ppk = long-term performance (як process реально performs): includes between-subgroup shifts + drift.

Empirically Ppk ≈ Cpk × 0.85 для typical stable process (1.5σ shift assumption — see Six Sigma section).

Threshold values (industry convention):

CpkInterpretationDefect rate (нормальний розподіл)
< 1.0Inadequate — process produces defects> 2 700 ppm
1.00Marginally capable2 700 ppm
1.33Capable (industry minimum)63 ppm
1.67Capable (preferred, automotive)0.57 ppm
2.00Six Sigma capability0.0019 ppm (з 1.5σ shift → 3.4 DPMO)

Automotive PPAP requirement: initial process study Pp ≥ 1.67 + Ppk ≥ 1.67 для special characteristics; Pp ≥ 1.33 + Ppk ≥ 1.33 для regular characteristics. Якщо process не досягає — submission rejected або interim approval з containment plan.

Cpm — Taguchi index (target-sensitive):

$$C_{pm} = \frac{C_p}{\sqrt{1 + \left(\frac{\mu - T}{\sigma}\right)^2}}$$

де T = target value. Cpm penalizes deviation from target additionally to deviation from spec — Taguchi’s loss-function philosophy.

9. MSA — Measurement System Analysis

MSA — Measurement System AnalysisAIAG MSA Reference Manual, 4th edition, 2010. Centrаl insight: measurement is also a process, з власним variation. Якщо ваше measurement variation еквівалентне process variation, ви не можете distinguish good parts from bad — ви фактично “вимірюєте noise”.

П’ять properties measurement system:

  1. Bias — systematic offset (measurement mean vs reference value).
  2. Linearity — consistency of bias across measurement range.
  3. Stability — consistency over time (drift).
  4. Repeatability — variation within same operator + same gage + same part (equipment variation, EV).
  5. Reproducibility — variation between operators (appraiser variation, AV).

Gage R&R = Repeatability + Reproducibility:

$$\sigma_{RR}^2 = \sigma_{EV}^2 + \sigma_{AV}^2$$

ANOVA method (preferred over older “Range method”): 2- або 3-factor crossed ANOVA з parts + operators + replicates як factors, partitions total variance into part-to-part + EV + AV + interaction components.

GRR % acceptance criteria (% of total study variation OR % of tolerance):

GRR %Verdict
< 10%Acceptable
10–30%Conditionally acceptable (consider cost of improvement vs criticality)
> 30%Unacceptable — measurement system inadequate

NDC (Number of Distinct Categories) — additional criterion:

$$NDC = 1.41 \cdot \frac{\sigma_{part}}{\sigma_{RR}}$$

NDC ≥ 5 required. NDC = 2 means measurement system can only distinguish 2 levels (essentially go/no-go); NDC ≥ 5 means it can resolve 5+ distinguishable levels within process variation.

Type-1 Gage Study (Cg / Cgk) — single-operator initial gage assessment перед full Gage R&R:

$$C_g = \frac{0.20 \cdot tolerance}{6 \cdot \sigma_{repeat}}, \quad C_{gk} = \frac{0.10 \cdot tolerance - |bias|}{3 \cdot \sigma_{repeat}}$$

Cg ≥ 1.33 + Cgk ≥ 1.33 = gage capable for that characteristic. Type-1 is prerequisite for full Gage R&R.

Attribute MSA (для pass/fail data) — uses Cohen’s Kappa statistic:

$$\kappa = \frac{p_o - p_e}{1 - p_e}$$

де p_o = observed agreement, p_e = expected agreement by chance. Kappa ≥ 0.75 = acceptable agreement; < 0.40 = poor agreement.

10. 8D — Eight Disciplines problem-solving

Ford Motor Company опублікував TOPS — Team Oriented Problem Solving у 1987 як formal methodology для multi-disciplinary problem solving. Хоча technique originated as 8 disciplines, today some industries include D0 як prep step, making “8D” actually 9 disciplines.

ДисциплінаНазваЗміст
D0Prepare and Emergency ResponsePlan + emergency response actions
D1Use a TeamCross-functional team з product/process knowledge
D2Describe the ProblemSpecify problem identifying 5W2H (who, what, where, when, why, how, how many)
D3Interim Containment ActionIsolate problem from customer (sort + segregate + temporary inspection)
D4Identify Root Causes + Escape PointAll possible causes + why detection failed
D5Verify Permanent Corrective ActionsConfirm chosen actions will resolve problem
D6Implement and Validate Permanent Corrective ActionsImplement + measure effect with empirical data
D7Prevent RecurrenceModify management + operation + practices + procedures
D8Recognize Team and Individual ContributionsFormal recognition

Ключова distinction: root cause vs escape point:

  • Root cause — fundamental reason problem occurred (Why is it broken?).
  • Escape point — control point in system that should have detected the problem but didn’t (Why didn’t we catch it before customer?).

Two independent corrective actions: (1) eliminate root cause, (2) improve detection. Naprykład: brake-pad bonding cure-temperature out of spec → root cause = controller PID tuning drift; escape point = SPC chart for cure-temp not monitored on weekend shift → fix both.

Tools used at each step:

  • D4 (RCA): 5 Whys (Toyota); Ishikawa fishbone diagram (Kaoru Ishikawa 1968, 6M categories: Manpower, Machine, Material, Method, Measurement, Mother Nature/Environment); Is/Is-Not analysis (Kepner-Tregoe); Pareto chart (80/20 — 80% of effects from 20% of causes); Fault Tree Analysis for safety-critical.
  • D5 (Verify): DOE (Design of Experiments); Monte Carlo simulation; pilot run з SPC monitoring.
  • D7 (Prevent recurrence): PFMEA update; control plan revision; lessons learned database.

8D report is formal customer-facing document — automotive OEMs require 8D submission after customer-reported nonconformity (standard 24-h/48-h/15-day submission cadence для D0 + D3 + D8 milestones).

11. Lean Manufacturing + Toyota Production System

Toyota Production System (TPS) розроблений у Toyota між 1948 і 1975 — ключові architects Sakichi Toyoda (loom autonomation, 1924), Kiichiro Toyoda (founder, JIT concept), Eiji Toyoda + Taiichi Ohno (codification post-WWII). TPS став основою Lean Manufacturing (Western terminology, popularized by Womack + Jones The Machine That Changed The World, 1990).

Дві pillars TPS:

  1. Jidoka — Automation with a Human Touch — machines auto-detect abnormality + stop themselves; operator не “babysits”. Походить від Sakichi Toyoda’s auto-stop loom (1924). Concrete implementation: Andon cord — будь-який оператор має authority + obligation stop the line при abnormality.
  2. Just-in-Time (JIT) — produce only what is needed, only when it is needed, only in the amount that is needed. Eliminates inventory waste. Implemented через Kanban (pull-signal cards / electronic equivalents) and Heijunka (production leveling — produce small lots of varying products in repeating cycle, not large batches).

Three problems TPS targets:

  • Muda (無駄, waste) — non-value-adding activity.
  • Mura (斑, unevenness) — variation у workload / output.
  • Muri (無理, overburden) — overload of people / machines.

Сім видів muda (Ohno’s original list, expanded to 8 in Western Lean):

#Waste (EN)Waste (UK)E-scooter manufacturing example
1TransportТранспортMoving battery cells across plant for grading + tabbing + welding sequentially without flow
2InventoryЗапаси30-day raw motor stator stock — capital tied up + obsolescence risk
3MotionРухOperator reaches across bench to grab fastener — fatigue + cycle time loss
4WaitingОчікуванняWelder idle while curing oven processes prior batch
5OverproductionПеревиробництвоBuilding 200 controllers when order is 150 (worst waste — generates inventory + transport + motion downstream)
6OverprocessingНадмірна обробкаPainting frame to mirror finish coverable area when matte black is sufficient
7DefectsДефектиRework + scrap + warranty claims
8Unused talentНевикористаний талантOperator who sees waste daily but has no Kaizen channel to suggest fix

Practical TPS tools (subset relevant to e-scooter manufacturing):

  • Kanban — pull signal: downstream consumer pulls from upstream provider as needed. Replaces push (build-to-schedule).
  • Heijunka — production leveling box / schedule: alternate models on assembly line (instead of batch 100 of model A then batch 100 of model B → alternate ABABAB).
  • Gemba — “the actual place”; managers go to factory floor + observe directly. Genchi Genbutsu (“go and see”).
  • Hansei — reflection + self-criticism after each project / event.
  • Kaizen — continuous improvement з small, frequent changes (vs Western “innovation = big leap” mentality). PDCA cycle (Plan-Do-Check-Act, Deming/Shewhart cycle).
  • 5S — workplace organization: Seiri (Sort) + Seiton (Set in order) + Seiso (Shine) + Seiketsu (Standardize) + Shitsuke (Sustain).
  • SMED — Single Minute Exchange of Dies (Shingo) — reduce tool changeover time to single-digit minutes; enables small-batch + JIT.
  • TPM — Total Productive Maintenance — operators perform basic maintenance + tracking, не лише dedicated maintenance team. Metric: OEE — Overall Equipment Effectiveness = Availability × Performance × Quality (world-class threshold ~85%).
  • Value Stream Mapping (VSM) — diagrams material + information flow з value-add vs non-value-add timing.
  • Hoshin Kanri — strategic policy deployment (top-down direction + bottom-up alignment).

12. Six Sigma — DMAIC + DMADV

Six Sigma запроваджено Bill Smith у Motorola у 1986 як statistical methodology to reduce defects. Jack Welch прийняв у GE у 1995, де воно стало centerpiece strategy і ~2/3 Fortune 500 компаній adopted to late 1990s.

Назва “Six Sigma” походить від statistical goal: ±6σ from process mean fits within specification limits3.4 defects per million opportunities (DPMO) — assuming 1.5σ long-term shift (process mean drifts ±1.5σ over time, so short-term ±6σ becomes effective ±4.5σ to nearest spec limit → 3.4 DPMO).

σ levelDPMO (з 1.5σ shift)Yield %
691 46230.85%
308 53869.15%
66 80793.32%
6 21099.38%
23399.977%
3.499.99966%

Дві improvement cycles:

  • DMAIC — for existing process improvement:
    • Define — project charter + scope + Voice of Customer (VoC) + Critical-to-Quality (CTQ) characteristics.
    • Measure — baseline performance + MSA + capability + sigma level.
    • Analyze — root-cause analysis з statistical tools (hypothesis testing, ANOVA, regression).
    • Improve — Design of Experiments (DoE) + pilot + verify.
    • Control — control plan + SPC monitoring + ongoing capability tracking.
  • DMADV / DFSS (Design for Six Sigma) — for new process / product design:
    • Define — design goals aligned to customer demands.
    • Measure — CTQs + measurement plan.
    • Analyze — design alternatives + concept selection.
    • Design — optimized solution з robust design (Taguchi methods).
    • Verify — pilot testing + validation.

Belt hierarchy (martial-arts inspired):

  • White / Yellow Belt — basic awareness, 1-2 days training.
  • Green Belt — part-time practitioner, leads small projects, ~1 week training.
  • Black Belt — full-time specialist, leads larger projects, 3-4 weeks training + certified project.
  • Master Black Belt — coach + mentor + portfolio leader.
  • Champion / Sponsor — executive sponsor + resource provider.

Key statistical tools Six Sigma practitioner uses: SPC + capability indices (sections 7+8), MSA (section 9), hypothesis testing (t-test, ANOVA, chi-square), regression, DoE (full factorial + fractional factorial + response surface methodology), Monte Carlo simulation.

Lean Six Sigma = TPS waste-elimination + Six Sigma statistical defect-reduction. Synergistic — TPS targets speed + flow, Six Sigma targets variation + accuracy. Together: fast and accurate.

13. Poka-yoke — mistake-proofing

Poka-yoke (ポカヨケ) — Japanese for “mistake-proofing” — формалізована Shigeo Shingo у Toyota у 1960s. Originally baka-yoke (“fool-proofing”) але renamed ~1963 за respect to workers. Shingo’s book Zero Quality Control: Source Inspection and the Poka-Yoke System (1986, English translation) є canonical reference.

Дві types:

  1. Warning poka-yoke — alerts operator that an error is about to occur (light / sound / vibration). Operator can still proceed if intentional.
  2. Control poka-yoke — physically prevents error from occurring at all. Operator cannot proceed if mistake is being made.

Три detection methods (Shingo’s classification):

  • Contact method — examines physical attributes (shape, dimension, color, position).
  • Fixed-value method — ensures correct count of motions / parts / operations (e.g., torque-tool counter that locks if not enough fasteners installed).
  • Motion-step method — verifies correct sequence completion (e.g., assembly software won’t allow Step 3 button until Step 2 is recorded complete).

Six principles (later expansion):

  • Elimination — change design so error is impossible (e.g., merge two parts so they can’t be assembled wrong).
  • Replacement — replace error-prone process with safer one (e.g., screw-driver with torque-control replacing manual).
  • Facilitation — make correct action easier than wrong (e.g., color-coded wiring harness connectors).
  • Detection — detect error after it occurs but before consequences propagate.
  • Mitigation — minimize impact when error does occur.
  • Prevention (also termed) — prevent error from being possible.

E-scooter manufacturing examples:

  • Battery connector polarity — asymmetric plug geometry (can only insert one way) is control poka-yoke / elimination.
  • Battery cell tabbing fixture — vision system rejects part if cell orientation wrong before welding — control poka-yoke / detection at source.
  • Brake-line bleeder valve — color-coded cap (red = open, green = closed) — warning poka-yoke / facilitation.
  • Fastener torque tool — locks after exceeding spec → cannot over-torque — control poka-yoke.
  • Wiring harness color-coding — phase A red, phase B yellow, phase C blue + connector shape — facilitation.
  • Folded-bike interlock — speed limiter активний until folding lever in locked position — control poka-yoke.
  • PCB orientation slot + key — board can only insert one way — elimination.

Poka-yoke є most cost-effective quality intervention — designed once into product / process, eliminates entire failure mode without ongoing inspection cost. SPC + Gage R&R cost recurring; poka-yoke amortizes once.

14. Cross-axis matrix — manufacturing-quality relevance до 30 попередніх axes

Engineering axis (попередня)Manufacturing-quality concept (це axis additionally constrains)
DT Joining (fastener torque)SPC X̄-R chart on torque tool output; Cpk ≥ 1.67 for safety-critical joints
DV Heat-dissipationThermal-paste thickness Gage R&R; cure-temp uniformity SPC
DX EMC/EMIShielding effectiveness 100% audit; ferrite-bead placement poka-yoke fixture
DZ CybersecurityProvisioning workflow: each unit gets unique key (poka-yoke = workflow can’t proceed without key burned); 100% read-back verification
EB NVHBearing pre-load Cpk ≥ 1.67; motor balance ISO 1940 G6.3 100% test
ED Functional safetySafety-critical characteristic per IATF 16949 8.3.3.3; 100% inspection + traceability per ISO 26262-7
EF SustainabilityRecyclable-material content batch tracking; ROHS / REACH compliance certificates per supplier PPAP
EH RepairabilityService-tool compatibility validated в DV + PV phases; spare-part part-number traceability
EJ Environmental conditioningIPX rating 100% production test; thermal-cycle ALT sample plan per AIAG SPC
EL PrivacySoftware image hash verified each unit; key burn-in poka-yoke (section 13)
EN ReliabilityFMEA → PFMEA → control plan chain (sections 4 + 6 + 7) is exactly the reliability engineering link
EP SW-processSoftware image release passes PPAP element 11 (initial process study на bootloader + factory provisioning); embedded software per IATF 8.4.2.3.1
ER Human factorsOperator station ergonomics (handles + lighting + reach) per ISO 14738; HMI poka-yoke for assembly errors
Battery / BMSCell capacity Cpk ≥ 1.67 (target 1.50 ±0.05 Ah → σ ≤ 0.005 Ah); IR matching ±5% within pack; cell-grade poka-yoke fixture
Brake systemPad friction μ Gage R&R on dynamometer; piston-stroke 100% functional test
Motor + controllerStator winding turn-count automated optical inspection (AOI); hi-pot test 1500 V 100% acceptance
SuspensionSpring rate Gage R&R; damper-fluid fill volume Cpk ≥ 2.0
TireCompound durometer (Shore A) SPC; tread depth 100% gauge
LightingLED bin sorting (luminous flux Cp ≥ 2.0); CRI batch QC
Frame + forkWeld penetration X-ray inspection 100% safety joints; yield strength batch certificate per PPAP element 10
HMI / displayPixel-defect AOI; backlight uniformity (corner-vs-center ratio) Cpk
ChargerOutput voltage Cpk ≥ 1.67; isolation hi-pot 100%; protection-trip burn-in test
Connector + harnessPull-test sample-plan AQL 0.65; continuity 100% automated; color-code poka-yoke
IP protectionSubmersion test sample plan; gasket compression Cpk
BearingInternal clearance Gage R&R; preload torque SPC; ISO 281 L10 batch consistency
Stem + foldingLatch-engagement force Cpk; folding cycle 100 000 ALT sample plan
DeckSandpaper-grit friction-coefficient Gage R&R; weight-rated proof-load 100% sampling
Handgrip + lever + throttleGrip-pull-off force AQL 1.0; throttle return-spring force Cpk
Wheel + rimSpoke-tension distribution Cpk; rim runout 100% indicator
Fastener (joint)(Same as DT — duplicate row to confirm axis-by-axis closure)

Кожна попередня axis отримує manufacturing-quality constraint як production-condition свого own design decision (e.g., battery cell axis designs cell chemistry to deliver target capacity, BUT manufacturing-quality constrains cell-to-cell variation Cpk and IR matching tolerance що feeds back to required upstream cell-grading + sorting protocol).

15. Owner-level manufacturing-quality “tells” — DIY checklist

8-step DIY manufacturing-quality assessment при отриманні нового e-scooter (or used + suspected of poor build):

  1. Batch serial cross-check — VIN / S/N + battery S/N + motor S/N + controller S/N: чи всі consistent date-codes (within 30 days)? Mixed date-codes може signal warranty replacement / refurb / mismatched components.
  2. Weld bead consistency — frame welds: чи bead width uniform along seam (Cpk-style visual proxy)? Uneven beads = manual welding without fixture / multiple welders / process out of control.
  3. Fastener torque marks — many factories mark torqued bolts з paint stripe (single line through bolt+nut+ground). Mark broken across line = bolt has been disturbed since factory. Marks completely absent = factory без torque-control discipline.
  4. Label-to-spec match — battery pack capacity label (e.g., “48V 20Ah”) matches actual measured capacity (run-time × current draw ≈ rated)? Off by > 10% = either bin grading bypass або low-capacity cell substitution.
  5. Paint / cosmetic AOI proxy — orange-peel, fish-eye, dust inclusion у paint? Factory без AOI line will show inconsistent finish across units. Compare two units of same model — variation between units > variation within single unit signals process not in control.
  6. PCB inspection — open the controller housing (if warranty-friendly): solder joints uniform, no cold joints / bridges / unflushed flux / damaged components? Hand-soldered PCB (uneven solder fillets) means no wave / reflow + AOI line — high probability of escape defects.
  7. Connector / harness color-coding — wires color-coded per industry convention (phase A red, B yellow, C blue для BLDC; +/- per battery convention)? Random colors = no poka-yoke design.
  8. Service manual + parts traceability — manufacturer publishes service manual з part numbers + torque specs + replacement procedures? If absent — factory has not invested in DV/PV documentation → likely also missing control plan + PFMEA discipline.

Owner-level “yellow flag” indicators:

  • Multiple identical units of same model show between-unit variation > expected (paint shade, hardware finish, label position). Healthy factory: ≤ 5% visible cross-unit variation.
  • Date code spread within single unit > 90 days suggests inventory carry / lot mixing.
  • Manufacturer responds to warranty claim з vague “we’ll replace the part” without root-cause analysis = no 8D culture.
  • Recall history — public recall database (NHTSA in US, RAPEX in EU) shows pattern of similar issues across model line = systemic manufacturing-quality issue.

Green flags:

  • Public ISO 9001:2015 / IATF 16949:2016 certificate from accredited body (verify через certifier’s website, not just claim on box).
  • Published warranty terms з clear 8D-style RMA process.
  • Spare parts available individually з part numbers + diagrams.
  • Service manual published з torque values + procedure detail.

16. Future axes — куди axis-серія розширюватиметься

Як reliability (EN), SW-process (EP), ergonomics (ER), і manufacturing-quality (ET), наступні process meta-axes:

  • Risk management (ISO 31000:2018 + ISO/IEC 31010:2019 + Bowtie + ALARP + LOPA) — risk-meta-axis верх HARA + TARA + reliability FMEA + manufacturing FMEA.
  • V&V engineering як standalone axis (IEEE 1012:2016 System, Software, and Hardware Verification and Validation) — поки разділене між functional-safety (ED), SW-process (EP), і manufacturing-quality (ET, PPAP V&V scope); IEEE 1012 окремий стандарт.
  • Production logistics & supply chain (ISO 28000:2022 Security and resilience — Security management systems + C-TPAT + AEO + UFLPA compliance) — flow axis.
  • Configuration management (ISO 10007:2017 Quality management — Guidelines for configuration management) — baseline + change-control axis.
  • Project management (ISO 21500:2021 + PMBOK + PRINCE2) — schedule/budget/scope axis.

Жодна з них не є prerequisite до manufacturing-quality-axis — порядок publication лишається на judgement автора, з основним критерієм «що зараз найбільш цінне для е-самокат power-user».

17. Reuse — manufacturing-quality concept-як-pattern

Cross-cutting infrastructure axis pattern v14 — fourteen-instance set (joining DT + heat-dissipation DV + interference-mitigation DX + interconnect-trust DZ + acoustic-vibration-emission EB + safety-integrity ED + sustainability EF + repairability EH + environmental-conditioning EJ + privacy-preservation EL + reliability-prediction EN + SW-process EP + human-machine-fit ER + manufacturing-process ET).

Manufacturing-quality, як reliability + SW + ergonomics — methodology layered over all others rather than separate subsystem:

  • Reliability (EN) описала формальний апарат, щоб прогнозувати і валідувати надійність every попередньої axis.
  • SW-process (EP) описав формальний апарат, щоб будувати і доставляти firmware, що реалізує decisions кожної з 28 axes.
  • Ergonomics (ER) описала формальний апарат, щоб fit людину з кожною з 29 попередніх axes у статиці й русі.
  • Manufacturing-quality (ET) описує формальний апарат, щоб серійно виробляти конкретні exemplars кожної з 30 попередніх axes у такій кількості й якості, що statistical defect rate (DPPM) залишається в acceptable bound, і кожен customer отримує той самий product, що пройшов DV/PV gates.

Recap 10 points:

  1. Manufacturing quality ≠ design ≠ inspection — own scope, own metrics, own standards.
  2. ISO 9001:2015 + 10-clause Annex SL + 7 quality principles + risk-based thinking foundation.
  3. IATF 16949:2016 layered automotive QMS з ~140 додаткових requirements + customer-specific requirements; 3-year certification з annual surveillance.
  4. APQP 5 phases (Plan & Define → Product Design → Process Design → Validation → Launch) + Control Plan як ключовий output.
  5. PPAP 18-element submission + 5 submission levels (default Level 3) + Part Submission Warrant; required at new part / engineering change / process change / 12-month dormancy.
  6. AIAG-VDA FMEA Handbook 2019 7-step approach + Action Priority (AP) replaces RPN; Severity 9-10 = High AP automatic.
  7. SPC + 7 control charts + Western Electric / Nelson rules + rational subgrouping; common-cause vs special-cause distinction is fundamental.
  8. Capability indices: Cp / Cpk (short-term) vs Pp / Ppk (long-term); Cpk ≥ 1.33 capable / 1.67 preferred / 2.0 Six Sigma.
  9. MSA Gage R&R < 10% acceptable; NDC ≥ 5 required; Type-1 Cg/Cgk prerequisite.
  10. 8D (Ford TOPS 1987) — root cause + escape point dual analysis; 5W2H + 5-Why + Ishikawa + Pareto toolset.

ENG-first джерела (0 російських, 30+ official):

  • ISO 9001:2015 Quality management systems — Requirementsiso.org/standard/62085.html
  • ISO 9000:2015 Quality management systems — Fundamentals and vocabulary (7 quality principles definition) — iso.org/standard/45481.html
  • ISO 9004:2018 Quality management — Quality of an organization — Guidance to achieve sustained successiso.org/standard/70397.html
  • ISO 19011:2018 Guidelines for auditing management systemsiso.org/standard/70017.html
  • IATF 16949:2016 Quality management system requirements for automotive production and relevant service parts organizationsiatfglobaloversight.org/iatf-169492016
  • IATF 16949:2016 FAQs + Sanctioned Interpretations (SIs) — iatfglobaloversight.org/iatf-169492016/iatf-169492016-sis
  • IATF Customer-Specific Requirements directoryiatfglobaloversight.org/oem-requirements/customer-specific-requirements
  • AIAG Advanced Product Quality Planning (APQP) Reference Manual, 2nd ed., 2008 — aiag.org
  • AIAG Production Part Approval Process (PPAP) Reference Manual, 4th ed., 2006 — aiag.org
  • AIAG Statistical Process Control (SPC) Reference Manual, 2nd ed., 2005 — aiag.org
  • AIAG Measurement Systems Analysis (MSA) Reference Manual, 4th ed., 2010 — aiag.org
  • AIAG & VDA Failure Mode and Effects Analysis FMEA Handbook, 1st ed., June 2019 — aiag.org/quality/automotive-core-tools/fmea
  • VDA Band 6.3 Process Audit, 3rd ed., 2016 — vda-qmc.de
  • VDA Band 6.5 Product Audit, 3rd ed., 2020 — vda-qmc.de
  • ANSI/ASQ Z1.4-2003 (R2018) Sampling Procedures and Tables for Inspection by Attributesasq.org/quality-resources/z14-z19
  • Ford Motor Company Team Oriented Problem Solving (TOPS) — 8D Methodology, 1987 (proprietary).
  • W. A. Shewhart Economic Control of Quality of Manufactured Product, Van Nostrand, 1931 (reprinted ASQ 1980).
  • W. E. Deming Out of the Crisis, MIT Press, 1986 (reissued 2018).
  • W. E. Deming The New Economics for Industry, Government, Education, MIT Press, 2nd ed. 1994.
  • J. M. Juran Juran’s Quality Handbook: The Complete Guide to Performance Excellence, 7th ed., McGraw-Hill, 2017.
  • P. B. Crosby Quality Is Free: The Art of Making Quality Certain, McGraw-Hill, 1979.
  • D. J. Wheeler Understanding Statistical Process Control, 3rd ed., SPC Press, 2010.
  • D. J. Wheeler Advanced Topics in Statistical Process Control, 2nd ed., SPC Press, 2004.
  • Taiichi Ohno Toyota Production System: Beyond Large-Scale Production, Productivity Press, 1988 (English translation; Japanese 1978).
  • Shigeo Shingo Zero Quality Control: Source Inspection and the Poka-Yoke System, Productivity Press, 1986 (English; Japanese 1985).
  • Shigeo Shingo A Revolution in Manufacturing: The SMED System, Productivity Press, 1985.
  • J. Womack, D. T. Jones, D. Roos The Machine That Changed The World: The Story of Lean Production, Free Press, 1990 (reissued 2007).
  • J. Womack, D. T. Jones Lean Thinking: Banish Waste and Create Wealth in Your Corporation, Free Press, 2nd ed., 2003.
  • Mikel Harry, Richard Schroeder Six Sigma: The Breakthrough Management Strategy Revolutionizing the World’s Top Corporations, Currency/Doubleday, 2000.
  • Thomas Pyzdek, Paul Keller The Six Sigma Handbook, 5th ed., McGraw-Hill Education, 2018.
  • Mary Walton The Deming Management Method, Perigee Books, 1988.
  • IEEE 1012-2016 IEEE Standard for System, Software, and Hardware Verification and Validationstandards.ieee.org/standard/1012-2016.html