SystemManufacturing Intelligence

Production Bottleneck Analysis

Production bottleneck analysis that maps where each batch waits, names the step that limits the plant with its sum, and rehearses each fix before anyone signs.

See one plant's constraint found and moved ↓
Watch with transcript →
The problem

The longest pile of batches sits in front of QA release

A batch spends most of its life waiting: for a cleaned granulator, a coater, the QC lab, its batch record review. Each area reports its own backlog in its own unit, batches, records, samples or hours, so the tier meeting cannot compare them, and the OEE dashboard points at the machines because machines are what it measures.

Nobody computes the load on record review. The release queue grows a few batches a week, a client asks where its batch is after the due date has gone, and someone walks it through four systems by hand. The improvement project starts on a guess and is checked months later, if at all.

Built forPlant headsOpEx leadsQA operations managersProduction planners

The Problem Today

Each area counts its own backlog, so plants chase machines while the queue in front of quality review grows.

62%
drugs in shortage after manufacturing or quality problems, FDA published

Backlogs counted four ways

  • Step times
  • Lab queue
  • Review log
  • Holds
  • Machines get the blameOEE exists for machines and says nothing about paperwork
  • Due dates missed by surpriseA client asks where its batch is after the date has passed

The Solution

  1. The plant drawn as a flow mapEach batch a square in the queue it waits in, step by step
  2. The constraint named with its sumDemand over capacity on each step, each press line on its own products
  3. Fixes rehearsed before money is spentEach option shows where the constraint moves next
  4. The plant head signs, QA co-signsThe plan becomes owned actions, measured each week

Built-in controls

  • QA still approves each record before release
  • Nobody approves the plan they proposed
  • Plant systems are read only

What You Get

The plant takes on more client volume by fixing the step that really limits it, and knows where the limit goes next.

  • Room for more batchesHeadroom shown before any building or hire
  • Clients warned with a dateA late batch traced, with a forecast and its reason
  • One method in four plantsEach plant's constraint found by the same rules
Features

Production bottleneck analysis features, from flow map to signed plan

The OpEx lead, the production planner, the QA operations manager and the plant head work one oral solids plant through them, where batches pile up in front of QA release.

demoEach plant, client, person, batch and number shown here is sample data.

  1. 01Agents at work

    Agents map the plant from step times, the lab worklist and the review log

    • Each queue counted from the plant's records. Step times, the QC worklist, the record review log and quality holds are read together, so each area's backlog is counted the same way.
    • Steered in plain words mid-run. The OpEx lead adds the QC lab while the finder is still working, and its plan changes in view.
  2. 02Flow map

    Manufacturing bottleneck analysis on a flow map, each batch in its queue

    • Load on each step, line by line. Each press line carries only the products validated on it, so the load reflects what the plant can really run.
    • Record review loaded like a machine. Batch record review gets a load figure beside the equipment, and the plant sees it limits output before coating does.
  3. 03Constraint evidence

    The sum behind the constraint, with the review log as QA keeps it

    • Capacity worked out from the rota. Reviewers, productive hours and hours per record give the step's capacity, set against the batches reaching it each week.
    • Weeks of queue growth required. A step is named the constraint only when its queue has grown over time, so one bad week does not set the plan.
  4. 04Batch trace

    Batch cycle time traced step by step, each wait labeled by cause

    • Waits labeled from records. Each wait carries its cause from a record, and a wait the records cannot explain stays unlabeled.
    • A forecast before the due date. The planner sees when a late batch will be reviewed and can ask QA operations to take it next, with QA deciding.
  5. 05Review by exception

    Review by exception rehearsed: reviews a week and where the limit moves

    • Assumptions visible and editable. Hours per record, the share of clean records and the rota show their basis, so QA can challenge each one.
    • The next constraint shown. The plant sees that faster review moves the limit to coating before anyone commits to it.
  6. 06Saturday shift

    A second fix layered on the first, with the batches a year it frees

    • Options combined and replayed. A coating shift added to the review change shows the QC lab as the next limit, so the plan names what follows it.
    • Headroom in batches a year. The rehearsal states how many more batches the plant could take before the next limit, an answer business development can give a client.
  7. 07Challenger

    Objections from QA, clients, maintenance and the planner before signing

    • The agents' own slip caught. A batch on hold for a deviation was counted in the backlog a speed-up would clear, and the challenger took it out.
    • Each objection with an owner. QA's validation needs, the clients' notice, the maintenance window and the planner's line limit each go to a named person.
  8. 08Plan sign-off

    Plant head approval after reading the rehearsal, QA co-signature apart

    • Read before signing. The plant head opens each part of the rehearsal before the sign button works, and signs with a meaning and a reason.
    • Review changes need QA too. An item that changes record review waits for the QA operations manager's own signature.
  9. 09Actions

    Owned actions that run through change control before go-live

    • Approved by someone who wrote neither. The validation report and the change control are approved by a QA compliance lead who did not write them.
    • Effect measured weekly. Once an item goes live, the tracker measures whether the constraint really moved and brings it to the weekly review.

Four roles share the constraint review, and two of them sign the plan

The OpEx lead proposes, the plant head approves, QA operations co-signs what touches review, and the planner lives with the result.

Plant heads

Own output and on-time delivery, and sign the plan

  • One named constraint with its sum
  • Fixes rehearsed before money is spent
  • Headroom stated in batches a year
OpEx leads

Rehearse options and propose the plan

  • Assumptions to edit and replay
  • Where the constraint moves next
  • Objections listed before the plant head reads
QA operations managers

Run record review and co-sign any change to it

  • Review load counted like a machine's
  • Review-order requests answered with a reason code
  • No review change live without validation
Production planners

Sequence campaigns and answer where a batch is

  • Each late batch traced wait by wait
  • A review forecast before the due date
  • One request covering the past-due batches

No agent names a constraint without two weeks of queue growth behind it

The flow mapper hands to the finder, the finder to the tracer, and the rehearser's work goes through the challenger.

Flow mapper

Reads step times, the QC worklist, the review log and the holds, and counts each queue.

  • Edits no record
  • Counts no batch twice
Constraint finder

Works out demand over capacity on each step and queue growth over the window, then names the constraint with its sum.

  • Names no constraint without two weeks of queue evidence
Batch tracer

Splits the release queue by what each batch waits for, and traces one batch with each wait labeled.

  • Guesses no cause: an unexplained wait stays unlabeled
What-if rehearser

Plays options forward with visible, editable assumptions, and shows the next constraint and the headroom.

  • Applies no change
Plan challenger

Lists what the QA head, the clients, maintenance and the planner would say, and checks the agents' own drafts.

  • Approves nothing; each point is fixed or sent to a named person
Action tracker

Turns a signed plan into owned actions and measures the effect each week.

  • Closes no action: only its owner does

What the finder and the challenger turned up that a machine dashboard misses

The constraint sat in record review

QA release ran at 120% load with 41 batches waiting, while no machine step reached full load.

A held batch counted in the speed-up

The rehearsal counted a batch on hold for a deviation among those review by exception would clear; the challenger cut the backlog to 40.

A product that runs on one line

One client's growth needs a second qualified compression line whatever the plan does, because its product runs on a single line.

demoSample data, from the working solution.

More screens

The batches in work, the plan view and the weekly constraint review

After the plan: the map as it would look, the record the weekly review reads, and the same rules applied to the other plants.

Batches in work and what each waits forEach batch from dispensing to disposition with its step, wait, days and client due date; past-due batches go to QA in one request.
The map as the plan would leave itIn plan view the constraint moves to the QC lab, and the map says plainly that nothing is live until QA co-signs and the change control is approved.
The weekly constraint reviewEight weeks of the release queue, this week's numbers and the decisions on the record, for the plant's weekly review.
Four plants, one way to find the limitEach plant keeps its own steps and equipment, and its constraint is found by the same load, growth and evidence rules.
Rules, sources and signersThe near-the-limit load, evidence window and growth test, each source read only with its status, and who signs what.
People and agents on one recordEach read, steer, rehearsal and signature with who, why, signature meaning and session.

demoSample data, from the working solution.

Six read-only systems behind a production bottleneck analysis

Manufacturing execution systemStep start and end times for each batch, step and line.
Electronic batch recordThe review log, send-backs and cleaning records.
Laboratory information management systemThe release testing worklist and what each sample waits for.
Quality management systemHolds, deviations and change control.
Enterprise resource planningEach client's due date for each batch.
Maintenance system and shift rosterMaintenance plans for presses and coaters, and who works which shift.

What plant heads ask about production bottleneck analysis

What is production bottleneck analysis?

Production bottleneck analysis finds the step that limits how many batches a plant can finish, and what fixing it would free. Here agents run production bottleneck analysis from the plant's own step times, lab worklist, review log and holds, name the constraint with its sum, and rehearse fixes for the plant head to sign.

Why look beyond OEE to find the constraint?

OEE measures machines. Batches also wait for the QC lab and for batch record review, which sit outside any machine dashboard, so the load is worked out on those steps too, in the same terms.

Does review by exception thin out batch record review?

No. The quality unit still reviews and approves each record before release; review by exception changes how a reviewer looks. It goes live only after a validation report and a change control approved by someone who did not write them, and QA operations co-signs the plan.

Where does the constraint move after a fix?

The rehearser plays each option forward with visible assumptions and shows the next limit, so the plant sees whether a review change moves the constraint to coating, and a weekend shift moves it on to the QC lab, before spending anything.

Can a client be told when a late batch will be ready?

The batch tracer labels each wait and forecasts when the batch will be reviewed, and a request to review it next is accepted or declined by QA operations with a reason. It gives no ship date: release stays with the quality unit and the client.

Find your plant's
real constraint.

Bring a few weeks of step times, your review log and the lab worklist, and the agents will draw your flow map and name the step that limits it.