Selected project / Anonymised manufacturing case

The line was staffed for a full shift.
The equipment produced for roughly three hours.

An independent information layer turned an established automated line into an observable production system — exposing hidden time loss, bottlenecks, quality drift and avoidable operating cost.

See what the data revealed

Business problem

The equipment was automated. The operation was still largely invisible.

PLC values described machine states, laboratory results arrived later, and production reports were assembled separately. Management could see that output was below expectation, but not where the working day disappeared or which intervention would produce the greatest return.

The first important finding

Presence at the line was being mistaken for productive equipment time.

A complete staffed shift was reconstructed from independent signals and production events. The result changed the improvement conversation immediately.

REPRESENTATIVE FULL SHIFT8 hStaffed production window

≈ 3 h Productive equipment operation

≈ 5 h Identifiable opportunity gap

Material and process waitingDownstream blockageStops and recoveryQuality holds and coordination

The largest capacity opportunity was not higher machine speed.IT WAS RECOVERING LOST OPERATING TIME.

From symptoms to causes

The platform showed where flow broke — and what happened next.

BOTTLENECK VISIBILITY

Several constraints were limiting the line at different moments.

Time-aligned sensor and event data separated the true constraints from the stations that merely appeared busy. Queues, starvation and downstream blockage became visible as a connected flow problem.

FEEDmaterial

→

PROCESS Astarved

→

PROCESS Brunning

→

INSPECTIONblocked

LABOUR DESIGN

One permanent workstation was not required by the real process.

Cycle evidence showed that the task did not justify continuous staffing as a dedicated position. The work could be reorganised around actual demand, avoiding recurring labour cost without reducing production control.

BEFOREDedicated stationStaffed throughout the shift

→

AFTERDemand-based taskWork triggered by real process need

QUALITY IN REAL TIME

Defects became a live process signal instead of a delayed report.

Production conditions, material context and laboratory quality were joined into one trace. Emerging quality loss could be seen during production, allowing intervention before more material became waste.

DEFECT SIGNAL · LIVEINTERVENTION

PROCESS
ADJUSTED

Production startTimeCurrent

CYCLE OPTIMISATION

The cycle was improved by examining its phases, not only its total duration.

Sensor events separated load, build, hold, release and waiting time. Engineers could reduce non-value-adding intervals while protecting the physical conditions linked to acceptable quality.

BEFORELOADBUILDWAIT / HOLDRELEASE

OPTIMISEDLOADBUILDHOLDRELEASE

Critical process conditions preserved · non-value-adding time reduced

Business value delivered

Better decisions came from one shared version of production reality.

Recovered operating time

Stop causes and flow interruptions became measurable, prioritised and visible during the shift — supporting systematic reduction of avoidable downtime.

Avoided recurring labour cost

Evidence showed that one continuously staffed workplace was unnecessary, enabling the operating model to be redesigned around actual demand.

Reduced scrap exposure

Quality deviation became visible while production was still running, reducing the amount of material processed before corrective action.

Improved cycle economics

Phase-level analysis exposed waiting and inefficient timing, allowing cycle optimisation without compromising the conditions required for quality.

Delivered independently from machine control

A modern information layer around the existing line.

The original PLC and safety logic remained operationally independent. The new system observed the physical process, joined production and quality context, and delivered role-specific information without taking control of the machine.

PHYSICAL PROCESSIndustrial sensors · Encoders · Machine events

EDGE & DATALinux services · Python processing · PostgreSQL

OPERATIONAL CONTEXTMaterials · Laboratory · Production orders

PEOPLE & DECISIONSOEE · Reports · Alerts · Engineering analysis

SYSTEM ARCHITECTURE · SENSOR INTEGRATION · LINUX EDGE · PYTHON · POSTGRESQL · LABORATORY INTEGRATION · OEE

Project note

This case is intentionally anonymised. The operational findings and solution pattern are preserved; customer identity, product details, financial values and sensitive process parameters are not disclosed.

Find the hidden factory

How many productive hours are inside your staffed shift?

Start with one line and enough independent evidence to separate assumptions from real operating behaviour.

Discuss a production visibility pilot Return to case studies →