Jidoka: Safe Stopping at an Abnormality

Why it matters

Jidoka means that a machine or a worker stops the work when an abnormality is seen, making the problem visible. Automatic stopping is one application of this approach; a safe stop started by the worker serves the same purpose. The goal is to prevent a defect from being carried into the next operation and to allow the problem to be examined where it occurred. Quality at the source is broader than picking conforming parts at final inspection. The condition under which work counts as conforming, how a deviation will be recognized, and who does what when a deviation occurs are all defined in advance. In a work area where the authority to stop has been given but no support arrangement has been set up, the worker is left alone between the stop and production pressure. For this reason technical detection and the management response are designed together.

An abnormality is not only a defect visible to the eye. A part may look conforming and yet product assurance is broken if a required process condition cannot be verified. A skipped check, a wrong fixture, or a lost measurement result are different events; what they share is a shortage of acceptance evidence. Not every deviation requires stopping the whole plant in the same way. The boundary of the affected operation, the linked transfer and the product group is determined. But this boundary cannot be narrowed without evidence. A safe stop does not mean intervening directly in moving equipment; the safe stop and authority arrangement defined for the task is followed. This lesson does not replace an equipment intervention instruction; it builds the logic of the quality decision.

An escape is a part that crossed the defined acceptance boundary and was later found nonconforming. This boundary may be the next station, the finished-goods area or shipping; the same boundary is kept when rates are compared. A defective part caught and set aside inside the line is not written into the same result as a part that crossed the boundary. Also, several defects on one part do not increase the count of defective parts. The numerator of the escape rate is the escaped parts and the denominator is the total parts in the same scope. When the observation period and delayed reports are not explained, the escapes of the new period may simply not have appeared yet. Presenting the low rate that arrives early as a lasting improvement is then a mistake; the maturity condition of the records must be equalized.

Stoppage time alone is not quality success. Silencing a sensor or skipping a check can reduce stoppage while raising escapes. Conversely, making previously hidden deviations visible can raise stoppages in the first period. This rise is not automatically success either; the causes, the product boundary and the remedy work are examined. Production, quality and maintenance must be able to read the same event record. In the record, the observed symptom and the possible cause are kept apart. Later lessons will detail call times, detection verification, risk priority and release. Here the basic decision is not to let an unverified product move on for the sake of reducing lost minutes.

Authority to stop, technical intervention and product approval are different duties. Until support arrives the product is protected, and the restart evidence is tied to the record. At shift change the identity of segregated containers, open investigations and follow-up conditions are handed over. A machine running does not close the earlier uncertainty. A shared event record lets the next team use the same boundary and acceptance condition.

Formula