Eliminating scanner dropouts at a 1.4M sq ft stamping plant

Context

A Tier-1 automotive supplier had recently rolled out a new MES that depended on barcode scanners at every stage of the stamping and welding lines. Within weeks, line supervisors were reporting scanner re-scans on roughly one in three scans — enough to slow throughput measurably and enough to make production lose confidence in the MES rollout.

The problem

The plant's Wi-Fi network had been designed five years earlier for an environment with significantly less metal on the floor. New automation, new welding stations, and a re-laid forklift route had changed the RF environment. The OEM had recommended adding access points; an internal consultant had recommended switching vendors. Neither recommendation was backed by measurement data.

Our approach

• We pulled the last 18 months of help-desk tickets, sorted by location and time-of-day. The dropouts clustered around shift changes and along two specific aisles — not the locations the OEM had recommended new APs.
• We walked the plant during three shifts, observing actual operator workflow and scanner roaming behavior. The slowest scanner in the fleet (a forklift-mounted unit from 2018) was driving the roaming problem; faster handhelds were tolerating the network better.
• We ran a calibrated on-site survey with a reference radio matching the worst-performing scanner. The data confirmed the help-desk pattern: coverage gaps along the two aisles, plus three multipath-prone zones around the welding bays.
• We designed a remediation: 23 antenna repositions, 6 new APs (not 14, as the OEM had quoted), updated channel and power plans, and a controller config change that fixed an aggressive roaming threshold.
• We wrote the work as a vendor-neutral SOW. The plant's existing integrator did the install in two scheduled outages over a single weekend.

The outcome

Within two weeks of cutover the scanner re-scan rate dropped from 32% to under 1%. Line supervisors confirmed throughput recovery without further attention. The plant retained the methodology document and survey data, and used it as a baseline for two later expansions.

"They told us where it would fail, and where it wouldn't. The OEM had been telling us to add hardware everywhere. Wireless Phone Services told us to move six antennas and reconfigure the controller. They were right."

Measurable results