mfg-oee-analysis

asgard-ai-platform ｜商業與營運｜原文｜ MIT

把 OEE 拆成稼動、性能、良率三項，找出產線損失在哪

中文翻譯

---
name: "mfg-oee-analysis"
description: "把整體設備效率（OEE）拆成稼動率、性能率和良率來計算與診斷。使用者需要衡量產線效率、找出設備損失、對製造績效做標竿比較，或要為資本投資提出佐證時使用，就算他們說的是「為什麼我們的產出這麼低」、「機台稼動率報告」、「生產效率」或「我們損失了多少產能」也一樣。"
metadata:
  category: "WP-03 製造業"
  tags: ["manufacturing", "oee", "production-efficiency"]
---

# OEE 分析

## 框架

```
IRON LAW: OEE = Availability × Performance × Quality

OEE is a MULTIPLICATIVE metric. 90% × 90% × 90% = 72.9%, not 90%.
Each factor compounds the loss. World-class OEE is 85%+. Most plants
operate at 60-65%. Knowing the TOTAL is useless — you must decompose
to find which factor is dragging performance down.
```

### 三個因子

| 因子 | 公式 | 衡量的是 | 損失類別 |
|--------|---------|----------|----------------|
| **稼動率（Availability）** | 運轉時間 / 計畫生產時間 | 運轉時間 vs 停機時間 | 設備故障、換線、缺料 |
| **性能率（Performance）** | （理想週期時間 × 總產量）/ 運轉時間 | 實際速度 vs 設計速度 | 小停機、低速運轉、空轉 |
| **良率（Quality）** | 良品數 / 總產量 | 良率、一次通過的品質 | 瑕疵、重工、報廢、開機不良品 |

### 六大損失（對應到 OEE 因子）

| 損失 | OEE 因子 | 例子 |
|------|-----------|---------|
| 1. 設備故障 | 稼動率 | 機台故障、非計畫性維修 |
| 2. 設置與換線 | 稼動率 | 產品換線、換模、清潔 |
| 3. 空轉與小停機 | 性能率 | 感測器被擋住、清除卡料、小幅調整 |
| 4. 速度降低 | 性能率 | 因磨損或物料而低於額定速度運轉 |
| 5. 製程不良 | 良率 | 製程中的不良品、重工 |
| 6. 開機不良 | 良率 | 暖機期間的報廢、首件不合格 |

### 計算範例

```
Planned Production Time: 480 min (8-hour shift)
Downtime (breakdowns + changeover): 60 min
Run Time: 420 min

Ideal Cycle Time: 1 min/unit
Total Units Produced: 380

Good Units: 360
Defective Units: 20

Availability = 420 / 480 = 87.5%
Performance = (1 × 380) / 420 = 90.5%
Quality = 360 / 380 = 94.7%

OEE = 87.5% × 90.5% × 94.7% = 75.0%
```

### 診斷步驟

**Phase 1：計算 OEE**，每條產線／每台機器都算
**Phase 2：找出最弱的因子**（稼動率、性能率或良率）
**Phase 3：對該因子內的損失做柏拉圖分析**（哪一項具體損失最大？）
**Phase 4：對最大的損失做根因分析**（5 Whys、魚骨圖）
**Phase 5：改善**並重新量測

### 標竿

| OEE 水準 | 評等 | 典型情況 |
|-----------|--------|---------|
| > 85% | 世界級 | 頂尖製造商 |
| 60-85% | 一般 | 還有改善空間 |
| 40-60% | 偏低 | 損失很大，需要緊急處理 |
| < 40% | 危急 | 設備或製程根本壞掉 |

## 輸出格式

```markdown
# OEE Report: {Production Line}

## OEE Summary
| Factor | Value | Benchmark | Status |
|--------|-------|-----------|--------|
| Availability | {%} | >90% | 🟢/🟡/🔴 |
| Performance | {%} | >95% | 🟢/🟡/🔴 |
| Quality | {%} | >99% | 🟢/🟡/🔴 |
| **OEE** | **{%}** | **>85%** | 🟢/🟡/🔴 |

## Loss Breakdown
| Loss | Minutes Lost | % of Total Loss | Priority |
|------|-------------|----------------|---------|
| {loss type} | {min} | {%} | 1/2/3 |

## Root Cause (Top Loss)
{5 Whys or fishbone analysis}

## Improvement Plan
| Action | Target Impact | Timeline | Owner |
|--------|-------------|----------|-------|
| {action} | +{X%} OEE | {weeks} | {who} |
```

## 注意事項

- **OEE 是以單一機台為單位，不是整廠**：整廠平均的 OEE 會藏住一個事實：一台是 95%、另一台是 45%，平均出來是 70%。要個別分析。
- **計畫性停機要排除**：OEE 衡量的是相對於「計畫」生產時間的損失。排定的保養、沒有生產的班次、計畫性停工，都要從分母排除。
- **100% OEE 不是目標**：那代表零換線、零瑕疵、100% 的時間都以最高速度運轉。追求 100% 反而可能推高成本（例如永遠不做預防保養）。關鍵產線的目標訂 85% 以上就好。
- **資料蒐集才是真正的挑戰**：人工追蹤 OEE 不準。要投資自動化的資料蒐集（感測器、MES 整合），才能量得可靠。

## 參考資料

- TPM（全面生產保養）方法論請看 `references/tpm.md`
- 自動化的 OEE 資料蒐集請看 `references/oee-automation.md`

英文原文

---
name: "mfg-oee-analysis"
description: "Calculate and diagnose Overall Equipment Effectiveness (OEE) by decomposing into Availability, Performance, and Quality rates. Use this skill when the user needs to measure production line efficiency, identify equipment losses, benchmark manufacturing performance, or justify capital investment — even if they say 'why is our output low', 'machine utilization report', 'production efficiency', or 'how much capacity are we losing'."
metadata:
  category: "WP-03 製造業"
  tags: ["manufacturing", "oee", "production-efficiency"]
---

# OEE Analysis

## Framework

```
IRON LAW: OEE = Availability × Performance × Quality

OEE is a MULTIPLICATIVE metric. 90% × 90% × 90% = 72.9%, not 90%.
Each factor compounds the loss. World-class OEE is 85%+. Most plants
operate at 60-65%. Knowing the TOTAL is useless — you must decompose
to find which factor is dragging performance down.
```

### The Three Factors

| Factor | Formula | Measures | Loss Categories |
|--------|---------|----------|----------------|
| **Availability** | Run Time / Planned Production Time | Uptime vs downtime | Equipment failures, changeovers, material shortages |
| **Performance** | (Ideal Cycle Time × Total Count) / Run Time | Actual speed vs design speed | Minor stops, slow running, idling |
| **Quality** | Good Count / Total Count | Yield, first-pass quality | Defects, rework, scrap, startup rejects |

### Six Big Losses (mapped to OEE factors)

| Loss | OEE Factor | Example |
|------|-----------|---------|
| 1. Equipment failure | Availability | Machine breakdown, unplanned repair |
| 2. Setup & changeover | Availability | Product changeover, die change, cleaning |
| 3. Idling & minor stops | Performance | Sensor blockage, jam clearing, small adjustments |
| 4. Reduced speed | Performance | Running below rated speed due to wear or material |
| 5. Process defects | Quality | In-process rejects, rework |
| 6. Startup rejects | Quality | Scrap during warm-up, first-article failures |

### Calculation Example

```
Planned Production Time: 480 min (8-hour shift)
Downtime (breakdowns + changeover): 60 min
Run Time: 420 min

Ideal Cycle Time: 1 min/unit
Total Units Produced: 380

Good Units: 360
Defective Units: 20

Availability = 420 / 480 = 87.5%
Performance = (1 × 380) / 420 = 90.5%
Quality = 360 / 380 = 94.7%

OEE = 87.5% × 90.5% × 94.7% = 75.0%
```

### Diagnosis Steps

**Phase 1: Calculate OEE** for each production line/machine
**Phase 2: Identify the weakest factor** (Availability, Performance, or Quality)
**Phase 3: Pareto the losses** within that factor (which specific loss is biggest?)
**Phase 4: Root cause analysis** on the top loss (5 Whys, fishbone)
**Phase 5: Improve** and remeasure

### Benchmarks

| OEE Level | Rating | Typical |
|-----------|--------|---------|
| > 85% | World-class | Top manufacturers |
| 60-85% | Typical | Room for improvement |
| 40-60% | Low | Significant losses, urgent action needed |
| < 40% | Critical | Equipment or process fundamentally broken |

## Output Format

```markdown
# OEE Report: {Production Line}

## OEE Summary
| Factor | Value | Benchmark | Status |
|--------|-------|-----------|--------|
| Availability | {%} | >90% | 🟢/🟡/🔴 |
| Performance | {%} | >95% | 🟢/🟡/🔴 |
| Quality | {%} | >99% | 🟢/🟡/🔴 |
| **OEE** | **{%}** | **>85%** | 🟢/🟡/🔴 |

## Loss Breakdown
| Loss | Minutes Lost | % of Total Loss | Priority |
|------|-------------|----------------|---------|
| {loss type} | {min} | {%} | 1/2/3 |

## Root Cause (Top Loss)
{5 Whys or fishbone analysis}

## Improvement Plan
| Action | Target Impact | Timeline | Owner |
|--------|-------------|----------|-------|
| {action} | +{X%} OEE | {weeks} | {who} |
```

## Gotchas

- **OEE is per machine, not per plant**: Plant-level OEE averages hide that one machine at 95% and another at 45% average to 70%. Analyze individually.
- **Planned downtime is excluded**: OEE measures losses against PLANNED production time. Scheduled maintenance, no-production shifts, and planned shutdowns are excluded from the denominator.
- **100% OEE is not the goal**: It would mean zero changeovers, zero defects, running at max speed 100% of the time. Pursuing 100% can increase costs (e.g., never doing preventive maintenance). Target 85%+ for critical lines.
- **Data collection is the real challenge**: Manual OEE tracking is inaccurate. Invest in automated data collection (sensors, MES integration) for reliable measurement.

## References

- For TPM (Total Productive Maintenance) methodology, see `references/tpm.md`
- For automated OEE data collection, see `references/oee-automation.md`