← Engineering index

02 / Digital revamp for production

What did yesterday’s whiteboard erase?

I had supervised 24 cleanroom operators. The magnetic board told us what was running. By morning, the evidence of what had interrupted the previous shift had often been erased.

Magnetic board showing machine assignments and production status.Before · board
My operator interface for machine state, production work and service requests.After · operator screen
The original production board and an anonymized view of the operator software I built.

A shift board lost the history behind its totals.

The board showed the current job. After a handoff we could no longer tell when yesterday’s changeover began or how often a machine had stopped. I built the screens and database so the operator could record the event at the machine. Production and maintenance could see it next morning. I collaborated on the dashboards and taught colleagues to maintain the system.

The wider improvement effort doubled throughput over about 18 months. Several changes contributed. My evidence for the software is the working operator interface, the maintenance view and their use in daily huddles. An investment decision still needs to identify which recovered hour can produce another good unit.

Point of work

Start with the person requesting a changeover.

I built the operator interfaces and databases. An operator could request a changeover or raise a red tag at the machine. The event and its timing could survive the handoff to the next shift.

A long stop and six short ones can cost the same hours. They call for different fixes. A total at the end of the week hides which happened.

Original controls for changeover requests, red tags, material requests and shift handover.
Original production controls. Personnel, job and equipment identifiers are masked.

Read the floor / daily huddle

Where did the hours go?

I worked with colleagues on live dashboards and trained others to maintain the system. Operators, maintenance and managers could discuss the same record.

My original preventive-maintenance dashboard showing event counts, hours, scheduled activity and work by equipment.
Actual maintenance dashboard. It shows counts and hours by period or equipment. A count of work orders alone cannot say which job releases production capacity.
Inspect a second original production screen
Original lead production screen with current machine states and requests.
Production lead view · anonymized.

Capacity / one-shift example

Six short stops can outrank one long outage.

An illustrative comparison shows what the event record lets a team decide. The two machines have different jobs in the same eight-hour shift.

Suppose the line’s limiting station makes four good units an hour and loses six five-minute intervals. Recover the 30 minutes and the line can finish up to two more units, if customers need them and the line cannot catch up later. A 45-minute outage on a spare machine can cost zero finished units that shift when other capacity covers it.

What would change my choice?If the spare feeds an operation that runs out of material, its outage can become the constraint. I would check the buffer and finished output before approving a repair on the strength of a downtime ranking.

The example uses assumed rates. The before-and-after screens and plant account document the deployed system.  Original project and dashboard