Chapter 5｜Results & KPIs
The previous chapters covered machine connectivity, MES, production reporting, Cycle Time deviation alerts, GA production scheduling, and mobile warehouse operations. Once these functions are used on the floor, the important question is not how many new technologies were installed. The question is whether recurring production work became easier to manage.
This chapter therefore focuses on measurable operating indicators rather than a separate “AI result.”
5.1 OEE Improvement
After machine-status monitoring was introduced, downtime and Cycle Time deviations became easier to see without waiting for the daily report to be completed.
During the overall improvement period, OEE increased from approximately 60% to a range of about 75–85%. Actual values still vary with product mix, mold-change frequency, order volume, and equipment condition.
Cycle Time deviation alerts identify machines that differ from their historical baseline. If a product normally runs at about 30 seconds on a specific machine and current cycles remain near 35 seconds, the system highlights the difference.
The production team still investigates the reason. Cooling changes, robot waiting, mold condition, material, process adjustment, or other factors may be involved.
Compared with the internal pre-improvement baseline period, unplanned downtime was reduced by approximately 15 hours per month after changes in scheduling, downtime-reason management, and shop-floor response.
Machine utilization records during the same improvement period increased from about 68% to near 90%. Utilization and OEE are tracked separately because OEE also includes performance and quality factors.
5.2 Product Yield Improvement
Quality improvement was not produced by a single alert. The more important change was the ability to return from a defect record to the corresponding production history.
Once production reporting links the work order, time interval, and machine data, QC can check which machine and mold were used and review Cycle Time and other stored process data from the same period.
During the broader process and quality-improvement period, internal yield records increased from approximately 92% to approximately 97%.
The practical difference is that the abnormal period can be identified. Instead of relying only on operator memory, the team can first trace the work order and production time.
Compared with the internal pre-improvement baseline period, rework- and scrap-related cost decreased by approximately 40% after defect classification, process adjustment, mold improvement, and shop-floor quality management were combined.
5.3 Energy Use Improvement
Energy performance cannot be evaluated only from the monthly electricity bill. Without production context, it is difficult to separate the effect of output volume, idle time, machine efficiency, or product cycle time.
The practical approach is to compare machine operation, production time, and actual energy records on the same timeline.
With changes to machine scheduling, idle management, process improvement, and energy review, overall factory electricity use during the improvement period decreased by approximately 8–12%.
When machine electricity use is viewed together with production time, the factory can investigate whether higher consumption is caused by excessive idle time, different product cycle times, or unnecessary waiting created by the schedule.
Some individual months showed electricity-cost improvement of approximately 10–12%, but electricity cost is also influenced by production volume, working days, product mix, season, and tariff structure. It should not be presented as a fixed “system energy-saving rate.”
5.4 Scheduling Time Reduction
Scheduling was one of the easiest processes to compare before and after digitalization.
Previously, planners manually checked machines, molds, due dates, and existing orders. When an urgent order arrived, the existing schedule often had to be reviewed again.
GA first processes the defined constraints and produces an initial schedule. The planner then adjusts it according to actual production conditions instead of starting from a blank schedule.
The algorithm can handle due dates, job sequence, available machines, molds, and other quantified constraints. Mold repair, unexpected machine downtime, staffing, and special customer requests still require planner judgment.
Internal operation records show that some urgent-order insertions could reach an initial reschedule plus manual adjustment in approximately eight minutes.
The important point is not that “eight minutes guarantees the due date.” It is that a large amount of repetitive comparison can be completed quickly enough to give the planner a usable starting point.
5.5 Production Reporting and Warehouse Time Reduction
Production reporting and warehouse improvements were more straightforward: remove repeated data entry.
If the work order, item number, and machine already exist in the system, operators should not type them again. If a QR Code identifies a warehouse record, the system can reuse existing data.
After simplifying the RWD reporting interface, the time required for one shop-floor reporting operation decreased by more than 75%.
The purpose of reporting is not to make the operator enter more data. Operators only need to provide information the machine cannot know, such as manually judged defect quantities, abnormal causes, or necessary notes.
With QR Code identification and the responsive warehouse interface, existing item, document, and work-order data can be reused instead of being entered again in ERP.
Internal operation records showed approximately 70% improvement in processing efficiency for these tasks, while also reducing opportunities for manual item-number, quantity, or document-entry errors.
Across Sections 5.1 to 5.5, the changes are not the result of one software module. Machine connectivity makes states more visible, MES connects work orders with production history, scheduling reduces repetitive calculation, and RWD / QR Code functions reduce repeated data entry.
For Power Tech Mold, these KPIs are primarily useful as a baseline. When the factory changes scheduling, equipment, molds, or work methods later, the team can compare the result instead of relying on memory.