SystemManufacturing Intelligence

OEE & Downtime Analysis

OEE and downtime analysis that reads each stop on the line against the operator's own shift notes, adds up the losses one cause made, and gives that cause an owner.

Follow one bottle line's shift from the tape to an owner ↓
Watch with transcript →
The problem

Dozens of small stops coded unspecified, and the operators had written why

A packaging line logs its stops in the MES, and the operator picks a reason code on the line screen while the line is down and the queue backs up. The real reason goes into the shift logbook as the operator wrote it, and nobody joins the two.

So the weekly OEE chart shows a tall bar of unspecified small stops, slow running carries no code at all, and one worn part can show up as small stops, lost speed and rejects in three different buckets. A bar has no owner, so the cause comes back next week, sometimes on the next line.

Built forProduction supervisorsPackaging line managersMaintenance techniciansOperational excellence leads

The Problem Today

The reason a line stopped is written in the shift logbook, while the OEE chart counts a code picked with the line down.

Kept in separate systems

  • Stop log
  • Shift notes
  • Alarms
  • Rejects
  • A bar with no ownerUnspecified small stops return the next week
  • Notes nobody readsThe operator's words never reach the OEE review

The Solution

  1. The shift drawn as a tapeMachine states from the MES, each operator note clipped to its stop
  2. One cause, three lossesSmall stops, slow running and rejects that share a cause, added up
  3. Recode proposed, original keptThe supervisor proposes, and the old code stays visible, struck through
  4. Countersigned, with an ownerThe line manager opens sampled stops, signs and names who fixes it

Built-in controls

  • A recode is countersigned after sampled stops are read
  • The original reason code stays visible, never erased
  • Held product stays QA's decision, never the line's

What You Get

Lost line time gets a cause and an owner the next morning, with what the operators wrote as the evidence.

  • Operators' notes countTheir logbook lines become the evidence for a cause
  • Codes that match what happenedThe original code kept beside the new one
  • Line time back for client ordersHours one cause took, returned with released materials
Features

OEE and downtime analysis features, from one shift tape to a named owner

A production supervisor and a packaging line manager read a bottle line's two shifts, where a new pallet of caps began slipping in the capper chucks after lunch.

demoEvery line, lot, count and person on these screens is sample data.

  1. 01Shift tape

    Downtime analysis on a tape of the shift's machine states

    • Two shifts on one tape. Running, slow, small stop, stop, starved and blocked are drawn from the MES in order, so the supervisor sees when the line changed.
    • OEE written out as arithmetic. Availability, performance and quality show their own numbers on the signed definition, so anyone can recompute the day.
    • Six big losses linked to the tape. Clicking a loss lights its stretch of the shift, from unplanned stops to start-up rejects.
  2. 02Agents at work

    Downtime reason codes tested against alarms and the operators' notes

    • Notes matched by time, station and words. The note matcher ties each logbook entry to the stops and slow stretches it explains.
    • Alarms read beside the code. A stop coded unspecified that carries a capper alarm is flagged for what it really was.
    • Widened by the supervisor. A typed request such as reading the night shift as well changes the agents' plan where everyone can see it.
  3. 03Challenger

    A challenger checks the agents' count against the MES alarm log

    • Alarm count put right. The draft claimed every unspecified stop carried a capper alarm; three were stopped by hand, and the draft now says so.
    • Questions an auditor would ask. Whether the original code is kept and whether rejects are counted twice are answered before a person relies on the draft.
  4. 04One cause

    Six big losses added up for one cause: small stops, slow running, rejects

    • Losses that share a cause, summed once. Small stops, slow running after the cap change and torque rejects are added for one cause, and nothing is counted twice.
    • Evidence the operators wrote. Three logbook lines from the capper operator are the evidence, next to the alarm log, the torque printout and the cap lot label.
    • Held product named, not judged. The screen says QA holds the bottles capped with that lot and that their fate is QA's decision.
  5. 05Recode

    Recode proposed by the supervisor, with the original code kept

    • Original and new code side by side. Each stop shows the code entered and the proposed code, and the original is never overwritten.
    • Confidence on each stop. Stops stopped by hand carry medium confidence and are always in the sample the line manager opens.
    • No signing your own proposal. The supervisor proposes and signs, and a different person must countersign.
  6. 06Sampled stop

    Line manager opens each sampled stop before countersigning

    • Read, then marked as read. Each sampled stop shows its MES record and the shift notes from the same half hour before it can be marked read.
    • Countersigning stays locked. The countersignature waits until every stop in the sample has been opened, so nobody signs on a summary.
  7. 07Countersigned

    Recode countersigned, with the cause's owner named

    • An owner on every big loss. The line manager names who fixes the cause, here supplier quality for the closure.
    • Reopening clears both signatures. Changing a countersigned recode reopens it, so no signature stays on codes its signer did not see.
  8. 08Owned loss

    Unexplained loss moved to owned, with what stays honestly unknown

    • From unexplained to owned. The loss the cause made moves out of the unexplained bucket once the recode is countersigned.
    • Unknown stays unknown. Two morning stops with no alarm and no note keep no reason, and the day operator is named as the person to ask.
  9. 09Causes

    OEE improvement plan: each cause with its owner and action

    • Two causes still without an owner. The day's causes are listed with stops, minutes and bottles, and the ones nobody owns yet stand out.
    • The sibling line checked. The cause keeper proposes the same chuck check on the other bottle line that runs the same cap.

Operational excellence keeps one OEE definition across all four plants

Supervisors propose, line managers countersign and name owners, maintenance fixes, and operational excellence keeps one definition for every plant.

Production supervisors

Read yesterday's lines, chase unexplained stops and propose recodes

  • Each stop beside the operator's note
  • Losses that share a cause, added up
  • Recodes drafted with confidence
Packaging line managers

Own line output against the client schedule and countersign recodes

  • Sampled stops to open before signing
  • An owner named for each big loss
  • Options rehearsed for the open window
Maintenance technicians

Work the CMMS queue for cappers, sealers and labelers

  • The machine, part and start time
  • Work orders drafted from causes
  • Completion recorded in fixed fields
Operational excellence leads

Draft OEE definition changes and make sure each cause has an owner

  • One signed definition at every plant
  • Causes with no owner flagged
  • Sibling lines checked for the same cause

A looser countersign sample needs the OE lead, never its proposer

Production supervisor
Sees
The stops coded unspecified, their alarms and the operators' notes
Signs
The recode proposal, with the new code
Packaging line manager
Sees
Each sampled stop opened, the hand-stopped ones first
Signs
The countersignature, naming the cause's owner
Supplier quality specialist
Sees
The cap supplier's questions, drafted as fields
Signs
Sending the supplier request, then accepting its answer
Operational excellence lead
Sees
A proposed change that loosens the countersign sample
Signs
The looser sample, never one they proposed themselves
More screens

Released caps win back line hours now; the bigger fix waits for QA

Around the recode: what the line manager can still recover this week, how each line compares, and what went to the supplier and maintenance.

Options rehearsed for the rest of the weekThree options played forward over the open window, counting only bottles capped with released caps; the option that needs QA counts nothing until QA decides.
Each line on the same definitionEvery line and machine at four plants, with availability, performance, quality, OEE and the points still without a reason.
A request to the cap supplier in fixed fieldsMold cavities, knurl depth, skirt diameter and any change since the last lot, asked as fields the supplier answers in its portal.
Work orders drafted from causesChuck inserts, a sealer filter and a chute adjustment, each raised from a cause and completed by maintenance in fixed fields.
Each proposal, opening and countersignatureWho proposed, opened, countersigned or changed what, what each signature meant and in which session; the CSV export holds it all.
The signed OEE definitionPlanned time, small stop cutoff, what counts as good, the reason code list and the approvals; a new version needs the head of operations and the data steward.

demoSample data, from the working solution.

A slipping cap lot hid in three loss buckets at once

Three stops the alarm log did not carry

The draft said each unspecified stop had a capper alarm; three were stopped by hand to straighten caps.

One cause in three buckets

Small stops, slow running and torque rejects from the same chucks and closures, summed once.

Slow running with no code

Speed lost after restarts was read from the machine states as start-up ramps.

Stops nobody explained

Two morning stops stay without a reason rather than take a guessed one, with someone named to ask.

The note matcher drafts each recode, and no agent can apply one

They read the MES, the alarm log and the shift logbook without changing any of them.

State reader

Reads MES machine states and the stop and alarm logs, builds the tape and applies the signed definition.

  • Changes no MES record
  • Uses no definition that is not signed
Note matcher

Matches each shift note to stops and slow stretches by time, station and words, and drafts recodes with confidence.

  • Recodes nothing by itself
  • Reads no free text sent from outside
Loss calculator

Works the six big losses and adds the losses of one cause across small stops, speed and rejects.

  • Counts no loss twice
  • Changes no definition
Challenger

Plays the auditor and the line manager against the draft and checks the agents' own claims against the sources.

  • Approves nothing
  • Answers no product-impact question, which is QA's
Rehearsal agent

Plays options forward over the open window with each assumption shown and editable.

  • Chooses no option
Request drafter

Drafts structured requests to outside parties and reads replies through their fields only.

  • Accepts no reply
  • Lets no free text change a case
Cause keeper

Keeps the day's causes per line with an owner each, and proposes the same check on sibling lines.

  • Leaves owner choice to the line manager

Questions line managers ask about OEE and downtime analysis

What is downtime analysis?

Downtime analysis finds why a line lost time and output. Here each stop and slow stretch on the MES tape is read against the operator's own shift notes, losses that share a cause are added up, and a supervisor and a line manager sign the new reason code.

How is OEE calculated here?

As availability times performance times quality on the plant's signed OEE definition, written out beside the tape, with each of the six big losses linked to its stretch of the shift.

Does a recode erase the code the operator picked?

No. The original code stays visible, struck through. The supervisor proposes it, and the line manager countersigns after opening sampled stops.

Does it rule on bottles made while the caps slipped?

No. Held product is QA's call, and the assistant names the QA head instead of answering.

Does it work across lines and plants?

Yes. Each line and machine at each plant is computed on the same signed definition, and a cause found on one line is proposed as a check on its sibling lines.

Why are so many stops coded unspecified?

Because the operator picks a downtime reason code on the line screen while the line is down and the queue backs up, and writes the real reason in the shift logbook. Here each code is tested against the MES alarms and the operators' own notes, and a recode the supervisor proposes keeps the original code in view.

Bring us the bar called
minor stop, unspecified.

Send one line's MES stop log and a week of shift notes, and we will give each stop a sourced reason.