起源故事 · 从 Liebig 木桶到 Cpk
Origin Story · From Liebig's Bucket to Cpk
1850 年,德国农化学家 Justus von Liebig 在研究农田肥力时提出 最小因子律(Law of the Minimum):
植物的生长由最缺乏的那种营养元素决定,氮、磷、钾里只要缺一种,其他给再多也长不好。
后人把这条规律画成"木桶图"——能装多少水由最短的那块板决定。
1980 年代美国汽车业(Ford / Chrysler / GM 联合写的 QS-9000 + 后来的 AIAG SPC 手册)把这一思想搬进质量管理: 一个过程能力 Cpk 不是看你两侧距离的平均,而是 min(左距, 右距)—— 短板侧决定一切,长板侧再宽也救不了。从此 Cp 看潜力、Cpk 看现实,成了行业铁律。 In 1850 the German agricultural chemist Justus von Liebig, studying soil fertility, wrote down the Law of the Minimum: a plant grows only as well as its scarcest nutrient allows — short on nitrogen, phosphorus, or potassium, and the other two won't compensate. Later teachers redrew the idea as a wooden bucket — how much water it holds is set by the shortest plank.
In the 1980s the US auto industry (Ford, Chrysler and GM via QS-9000, then the AIAG SPC manual) ported the same logic into quality: process capability Cpk is not the average of the two distances, it is min(left, right) — the shorter side decides everything, the longer side cannot rescue it. From then on Cp measured potential and Cpk measured reality. The rule has not moved since.
1980 年代美国汽车业(Ford / Chrysler / GM 联合写的 QS-9000 + 后来的 AIAG SPC 手册)把这一思想搬进质量管理: 一个过程能力 Cpk 不是看你两侧距离的平均,而是 min(左距, 右距)—— 短板侧决定一切,长板侧再宽也救不了。从此 Cp 看潜力、Cpk 看现实,成了行业铁律。 In 1850 the German agricultural chemist Justus von Liebig, studying soil fertility, wrote down the Law of the Minimum: a plant grows only as well as its scarcest nutrient allows — short on nitrogen, phosphorus, or potassium, and the other two won't compensate. Later teachers redrew the idea as a wooden bucket — how much water it holds is set by the shortest plank.
In the 1980s the US auto industry (Ford, Chrysler and GM via QS-9000, then the AIAG SPC manual) ported the same logic into quality: process capability Cpk is not the average of the two distances, it is min(left, right) — the shorter side decides everything, the longer side cannot rescue it. From then on Cp measured potential and Cpk measured reality. The rule has not moved since.
1 双侧测量臂:哪边短哪边定 Cpk Two Measuring Arms: The Shorter One Sets Cpk
两侧相等tied2 木桶定律可视化:水位 = 短板高度 The Bucket Law in Pictures: Water Level = Shortest Plank
Liebig 1850两块木板:左板高 = 左距,右板高 = 右距。水位线 = min(两板),就是 Cpk。拖 μ 看哪块板缩短、水位跟着哪块走——其他板再高也没用。 Two planks: left plank = left distance, right plank = right distance. Water line = min(planks) — that's Cpk. Drag μ and watch the shorter plank drag the water down with it; the taller plank can do nothing about it.
3 现实里的短板 Shortest Planks in the Real World
跑道噪音 USL:噪音只有上限没有下限,单边规格 Cpk 退化成 (USL−μ)/(3σ),没有 min 可选。
Runway noise, USL only: noise has an upper limit but no lower one — single-sided spec, Cpk collapses to (USL − μ) / (3 σ). No min to take.
紧固件偏心 0.3mm:σ=0.1 看着精度很高,但左距=1.0、右距=2.0 → Cpk=1.0(不是平均的 1.5)。
Fastener offset 0.3 mm: σ = 0.1 looks tight, but left distance = 1.0, right distance = 2.0 → Cpk = 1.0 (not the average 1.5).
烤箱温度偏低 5°C:哪怕标准差极小,μ 离 LSL 太近,Cpk 立刻吃亏,半成品发硬。
Oven running 5 °C cold: even with razor-thin σ, μ is too close to LSL — Cpk drops at once, and half-baked goods come out hard.
客服响应 USL=60s:μ=45 σ=10,单边 Cpk=(60−45)/30=0.5——偏差太大要么提速要么放宽 USL。
Customer-service response, USL = 60 s: μ = 45 s, σ = 10 s → single-sided Cpk = (60 − 45) / 30 = 0.5. Either speed up or loosen the USL.
一句话In One Line
Cp 看潜力(钟胖瘦),Cpk 看现实(钟到最近规格的距离)。
偏心 + 高精度仍可能 Cpk 差,因为短板在偏的那一侧。
改进顺序也由此决定:
· 第一步:把 μ 调到中心——便宜,只需要调一次工艺设定、对一次模具,Cpk 立刻翻倍;
· 第二步:再降 σ——贵,要换设备、升级工艺、培训人员。
先做哪步,看左距和右距的差:差得越大,越说明偏心问题大过波动问题,先调 μ 性价比最高。 Cp tells you the potential (how lean the bell is). Cpk tells you reality (how close the bell sits to the nearest spec). High precision plus a small offset can still wreck Cpk — the shortest plank lives on the side you tilted toward.
And that fixes the order of improvement:
· Step 1 — recenter μ: cheap, one process-setting tweak or one tool offset, and Cpk often doubles overnight.
· Step 2 — then shrink σ: expensive, needs new equipment, process upgrades, operator training.
Which step first? Compare left vs. right distance. The bigger the gap, the more your problem is offset, not spread — recentering μ is the best ROI move.
改进顺序也由此决定:
· 第一步:把 μ 调到中心——便宜,只需要调一次工艺设定、对一次模具,Cpk 立刻翻倍;
· 第二步:再降 σ——贵,要换设备、升级工艺、培训人员。
先做哪步,看左距和右距的差:差得越大,越说明偏心问题大过波动问题,先调 μ 性价比最高。 Cp tells you the potential (how lean the bell is). Cpk tells you reality (how close the bell sits to the nearest spec). High precision plus a small offset can still wreck Cpk — the shortest plank lives on the side you tilted toward.
And that fixes the order of improvement:
· Step 1 — recenter μ: cheap, one process-setting tweak or one tool offset, and Cpk often doubles overnight.
· Step 2 — then shrink σ: expensive, needs new equipment, process upgrades, operator training.
Which step first? Compare left vs. right distance. The bigger the gap, the more your problem is offset, not spread — recentering μ is the best ROI move.
常见误用Common Mistakes
把 Cpk 算成 (USL−LSL)/(6σ)。那是 Cp(潜力),Cpk 必须包含 μ 的位置;少了 μ 就丢了"短板"信息。
Computing Cpk as (USL − LSL) / (6 σ). That's Cp (the potential). Cpk must include where μ sits — drop μ and you've thrown away the "short plank" signal.
单边规格也写 min。只有 USL(如噪音)或只有 LSL(如强度)时,Cpk 用单侧公式,没有 min 可取。
Forcing min onto a single-sided spec. With only a USL (noise) or only an LSL (strength), Cpk uses the one-sided formula — there's nothing for min to pick from.
报告 Cpk 不报 μ。Cpk 低必须分清:是偏心(先调 μ,便宜)还是波动大(降 σ,贵)——改进路径完全不同。
Reporting Cpk without μ alongside. A low Cpk has two very different causes — offset (recenter μ, cheap) or spread (shrink σ, expensive) — and the fix path depends entirely on which.