Figures come from project measurements, client-supplied solution documents, or public industry benchmarks.
A printing company
Legacy equipment connectivity and OEE improvement
The printing company's 54 legacy devices (PLCs, IPCs and I/O nodes) had siloed data, with downtime and plate-change causes recorded by hand and impossible to trace. We connected them at a 15-second sampling interval, built equipment dashboards, OEE analysis and parameter alarms, and structured downtime causes down to specific items such as machine collision and sheet tipping.
Measured single-machine OEE during the project was 70.73% (600 min scheduled time: 143.55 min planned downtime, 133.6 min unplanned downtime, 0 min defect time). Addressing downtime root causes cut unplanned downtime by 30–50% and lifted OEE by 8.8–14.6 percentage points.
An aluminium processor
Energy monitoring and optimisation
Energy was billed at workshop level with manual meter reading and no digital ledger, so high-consumption areas could not be pinpointed. We deployed 97 smart electricity and water meters for sub-item metering, real-time reading and loss allocation — every meter online — and established workshop- and asset-level baselines together with a unit-output energy cost KPI.
Energy labour down 50–60%, energy waste down 10% and O&M cost down 5%. Per GB/T 13234 (energy savings measurement and verification), a 10% cut in waste equates to roughly a 10% drop in energy per unit of output. Using the GB/T 2589 conversion factor (0.1229 kgce/kWh) and China's grid average emission factor published by the Ministry of Ecology and Environment (0.5703 tCO₂/MWh), every 1 million kWh saved equals 122.9 tce and avoids 570.3 tCO₂.
A retail chain
Store POS linked to ERP inventory in real time
POS data was consolidated T+1, so stock and sales never matched and oversell and stockouts were common. We unified POS and handheld terminals, aggregated transactions in real time through the middleware and wrote them into ERP, giving checkout, stocktake and transfers one set of definitions.
Stock-to-sales lag cut from T+1 to real time. Using published industry norms (mid-size supermarket turnover 8–15×/year, i.e. 24–45 days), we estimate real-time inventory can shorten turnover by 3–6 days (an estimate, not a project measurement); stocktake moves to scan and portal-reader capture with parallel multi-store counts and automatic reconciliation.
A footwear & apparel brand
Store transfers and warehouse picking integrated with ERP / WMS
With many style, size and colour combinations and frequent store transfers, head office, warehouse and store stock never reconciled. PDAs and barcode/RFID printers capture allocation, transfer, return and stocktake data, integrated with ERP and WMS through the middleware.
Stock accuracy rises from the industry-typical 70–85% with barcodes to over 99% (baseline is an industry range; results are project figures); store stocktake falls from 4–8 hours to minutes; shrinkage drops 30–50% (an estimate); head office, warehouse and store stock is reconciled from item-level movement records.
A pharmacy & beauty company
Batch and expiry management with end-to-end traceability
Batch numbers, expiry dates and onward flow were managed on paper, so a single batch recall took days to trace. Unique coding and batch binding happen on the line and in the warehouse, inbound and outbound scans leave a trail, and ERP/WMS integration delivers expiry alerts and flow control.
Batch traceability moves from paper ledgers taking days to in-system lookup, traceable down to dispatched orders; expiry thresholds are tiered by channel (online requires at least 1/3–1/2 of shelf life remaining), skincare alerts 6 months ahead, and picking follows FEFO.
An electronic circuit manufacturer
RFID warehouse management: automated inbound, outbound and put-away finding
Inbound and outbound relied on scanning items one by one, mis-shipments and stock discrepancies were common, and finding goods in the warehouse was slow. We deployed an RFID warehouse solution in a three-layer architecture: roller-door channels and gantry readers for bulk reading, PDAs for counting, printers for one-item-one-chip-one-code binding; the middleware talks to WMS / SAP in milliseconds and writes stock back in real time.
Inbound moved from item-by-item scanning to bulk channel reading — comparing about 3 seconds per item by scanner with a channel reading 300+ tags in under 10 seconds, roughly a 5× gain (an estimate); per device specification a roller-door channel reads 300+ tags within 10 seconds, with acquisition accuracy of 99% or above; outbound automatically reconciles goods against orders — green light to release, red light on discrepancy.
